Risk factors for the onset of panic and generalised anxiety disorders in the general adult population: A systematic review of cohort studies

Review

Risk factors for the onset of panic and generalised anxiety disordersin the general adult population: A systematic review of cohort studies

Patricia Moreno-Peral a,b,c, Sonia Conejo-Cerón a,b,c, Emma Motrico b,d,e,Alberto Rodríguez-Morejón b,c,f, Anna Fernández b,g,h, Javier García-Campayo b,i,Miquel Roca b,j, Antoni Serrano-Blanco b,h, Maria Rubio-Valera b,h,k,Juan Ángel Bellón a,b,c,l,m,n

a Unidad de Investigación, del Distrito Sanitario de Atención Primaria Málaga-Guadalhorce, Spainb Red de Investigación en Actividades Preventivas y Promoción de la Salud, ISCIII (redIAPP), Spainc Instituto de Investigación Biomédica de Málaga (IBIMA), Spaind Departamento de Psicología Evolutiva y de la Educación, Universidad de Sevilla, Spaine Departamento de Psicología, Sociología y Trabajo Social. Universidad Loyola Andalucía, Sevilla, Spainf Departamento de Personalidad, Evaluación y Tratamiento Psicológico. Universidad de Málaga, Spaing Faculty of Health Sciences, Centre for Disability Research and Policy, Brain and Mind Research Institute University of Sydney, Australiah Fundacio Sant Joan de Deu per a la Recerca i la Docencia, Parc Sanitari Sant Joan de Deu, Barcelona, Spaini Servicio de Psiquiatría, Hospital Miguel Servet, Instituto Aragonés Ciencias de la Salud, Zaragoza, Spainj Institut Universitari d’Investigació en Ciències de la Salut (IUNICS), Universidad de las Islas Baleares, Palma de Mallorca, Spaink School of Pharmacy, Universitat de Barcelona, Spainl Centro de Salud El Palo, Servicio Andaluz de Salud, Spainm Departamento de Medicina Preventiva y Salud Pública, Universidad de Málaga, Campus de Teatinos, 29071 Málaga, Spain

a r t i c l e i n f o

Article history:Received 12 March 2014Received in revised form9 June 2014Accepted 10 June 2014Available online 19 June 2014

Keywords:Anxiety disordersPanic disorderGeneralised anxiety disorderRisk factorsSystematic review

a b s t r a c t

Background: We aimed to assess available evidence on risk factors associated with the onset of panicdisorder (PD) and/or generalised anxiety disorder (GAD) in cohort studies in the general adult population.Methods: Systematic review using MEDLINE, PsycINFO and Embase. Search terms included panic disorder,generalised anxiety disorder, cohort studies and risk factors.Results: We finally selected 21 studies, involving 163,366 persons with a median follow-up of 5 years. 1)Sociodemographic factors: PD was associated with age, female gender, and few economic resources. GADwasassociated with age, non-Hispanics and Blacks, being divorced or widowed, and few economic resources. 2)Psychosocial factors: PD was associated with smoking and alcohol problems. GAD was associated withstressful life events in childhood and adulthood, and personality. 3) Physical and mental health factors: PDwas associated with the number of physical diseases suffered and the joint hypermobility syndrome. PD wasalso associated with a parental history of mental disorders, as well as with other anxiety disorders and othermental health problems in the person affected. GAD was associated with a parental history of mentaldisorders, as well as with other anxiety disorders and other mental health problems in the person affected,plus already having received psychiatric care.Limitations: Few studies examined the same risk factors.Conclusions: Sociodemographic, psychosocial and mental–physical health risk factors were determinant forthe onset of PD and GAD in the general adult population. These findings could be useful for developingpreventive interventions in PD and GAD.

& 2014 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3382. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jad

Journal of Affective Disorders

http://dx.doi.org/10.1016/j.jad.2014.06.0210165-0327/& 2014 Elsevier B.V. All rights reserved.

n Corresponding author at: Departamento de Medicina Preventiva Facultad de Medicina, Universidad de Málaga, Campus de Teatinos, 29071 Málaga, Spain.Tel.: !34 952 131515; fax: !34 952 137131.

E-mail address: [email protected] (J. Ángel Bellón).

Journal of Affective Disorders 168 (2014) 337–348

2.1. Search strategy and selection of articles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3392.2. Study selection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3392.3. Summary measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3392.4. Data synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3392.5. Assessment of risk of bias . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3393.1. Search results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3393.2. Risk of bias in individual studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3443.3. Sociodemographic risk factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344

3.3.1. Panic disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3443.3.2. Generalised anxiety disorder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344

3.4. Psychosocial risk factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3453.4.1. Panic disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3453.4.2. Generalised anxiety disorder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345

3.5. Physical and mental health risk factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3453.5.1. Panic disorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3453.5.2. Generalised anxiety disorder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346

4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3464.1. Main results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3464.2. Strengths and limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3464.3. Comparison with other systematic reviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346

4.3.1. Sociodemographic factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3464.3.2. Psychosocial factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3474.3.3. Physical and mental health factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347

4.4. Practical implications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3475. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347Conflict of interest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347Role of funding source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347

1. Introduction

Anxiety disorders represent a major public health problem. The12-month prevalence in the general population is 6–18% (Alonso etal., 2004; Andlin-Sobocki and Wittchen, 2005; Kessler et al., 2005).Two of the most common anxiety disorders are panic disorder (PD)and generalised anxiety disorder (GAD), with respective 12-monthprevalence rates in the general population of 0.8–2.7% and 1.0–3.7%(Alonso et al., 2004; Kessler et al., 2005; Somers et al., 2006).

Anxiety disorders are persistent and incapacitating (Kessleret al., 2005), and were the sixth leading cause of disability in 2010,in terms of years of life lived with disability, in both high-incomeand low- and middle-income countries (Baxter et al., 2014). Inaddition, PD and GAD are usually accompanied by other mentaldisorders (major depression and other anxiety disorders) (Kessleret al., 2005) and physical problems (cardiovascular, neurologicaland oncological diseases) (Gili et al., 2010; Smoller et al., 2007). Ofpatients with anxiety disorders, 50% have another comorbidpsychiatric disorder like depression or other anxiety disorder(Mergl et al., 2007).

Interest in developing programs to prevent anxiety disordershas been increasing (Feldner et al., 2004). The key to progressing

in the creation of these prevention programs resides in a clearunderstanding of the risk factors that predict the disorder(Schmidt and Zvolensky, 2007).

Most of the reviews so far undertaken have either focused onspecific populations of persons, such as those with epilepsy(Gandy et al., 2012), critical illness survivors (Davydow et al.,2009, 2010), children or adolescents (Beesdo-Baum and Knappe,2012; Beesdo et al., 2009), the elderly (Vink et al., 2008), andresidents in low-income countries and with particular social andhealth circumstances (Mirza and Jenkins, 2004; Tanios et al.,2009). Others have focused on particular risk factors, familyconditions or circumstances (Rapee, 2012), behavioural inhibition(Lahat et al., 2011), obesity (Gariepy, et al., 2010), smoking (Cosciet al., 2010) or early separation anxiety (Silove et al., 1996).However, only a few reviews have looked at a large number ofrisk factors in general populations. Furthermore, these reviewsgenerally possess a limitation concerning the design of the studiesincluded in the reviews. Cross-sectional studies cannot determinethat the risk factor precedes the effect (Mirza and Jenkins, 2004;Tanios et al., 2009; Vink et al., 2008), and although case-controlstudies do permit this, they are usually more biased than cohort

Table 1Inclusion and exclusion criteria for the studies included in the review.

Aspects considered Inclusion criteriaa Exclusion criteria

Population Z18 years at the end of the follow-up o18 years at the end of the follow-upDependent variable PD and/or GAD Diseases different to PD or GADIndependent variable Any risk factors except genes Genetic risk factorsPublication type Original studies Investigation protocols, clinical case, editorials and editor's letterLanguage All languages NoneDesign Prospective cohort with outcome not present at baseline Cross-sectional, case control, clinical trials, systematic reviews and meta-analysisSetting General population Clinical samples, secondary and tertiary healthcare, work and military setting

a PD: Panic disorder; GAD: Generalised anxiety disorder.

P. Moreno-Peral et al. / Journal of Affective Disorders 168 (2014) 337–348338

studies (Klauke et al., 2010). The present study, therefore, aimed todetermine the risk factors that can predict the onset of PD and/orGAD in cohort studies in the general adult population.

2. Methods

2.1. Search strategy and selection of articles

We followed the PRISMA guidelines for reporting systematicreviews (Moher et al., 2009). The protocol of this systematic reviewhas already been registered (International Prospective Register ofSystematic Reviews, registration number: CRD42013004121). Weperformed a systematic review of the published literature for cohortstudies aimed at identifying risk factors for the onset of PD and/orGAD in the general adult population. The databases were searchedseparately by 2 reviewers (PMP and SCC). The literature searcheswere completed from inception to September 2013 without anyrestrictions through the MEDLINE (PubMed), PsycINFO and Embasedatabases. The last search was run on 10 September 2013.

The PubMed search employed the clinical queries resource, whichlimits the search results to specific areas of clinical research (itexcludes basic research and reviews or meta-analyses). These searchfilters were designed by Haynes et al. (1994) and are widely acceptedamong the scientific community (Wong et al., 2003). The particularfilter used was etiology/narrow, the search strategy of which is:“((relative[Title/Abstract] AND risk*[Title/Abstract]) OR (relative risk[Text Word]) OR risks[Text Word] OR cohort studies[MeSH:noexp] OR(cohort[Title/Abstract] AND stud*[Title/Abstract]))”. In addition, usingdifferent combinations, the following key words were included: panicdisorder, generalized anxiety disorder, generalized anxiety disorderand risk factors. Finally, the search strategy used was: [(Etiology/Narrow[filter]) AND ((panic disorder OR generalized anxiety disorderOR generalized anxiety disorder) AND (risk factors OR risk factor)].Searches were piloted in PubMed then adapted to run across PsycINFOand Embase. The reference lists of the primary studies selected werealso checked.

2.2. Study selection

Studies were screened for inclusion over 3 phases. First,a search was made for any duplicates, which were then deleted;second, a selection was made based on the title and abstract; andthird, a final selection was made after reading the full text of thearticles. The whole selection process was done in duplicate (PMPand SCC), and in the event of any disagreement a third reviewer(ARM) was consulted in order to reach a consensus. The inter-agreement between PMP and SCC measured with the Kappastatistic was 0.81 (95% CI: 0.68–0.95). Fleiss (1981) characterisedkappas over 0.75 as excellent, 0.40 to 0.75 as fair to good, andbelow 0.40 as poor.

The articles were selected according to inclusion and exclusioncriteria specifically defined for this review (Table 1).

The inclusion criteria were considered for various reasons. Thereason for including persons older than 18 years of age at the end ofthe follow-up was because we aimed to determine risk factors forthe onset of GAD and PD in adults, regardless of the age at whichrisk factors may act. Moreover, there are already published reviewson the onset of anxiety disorders in children and adolescents(Beesdo-Baum and Knappe, 2012; Beesdo et al., 2009). We focusedon PD and GAD because they are two of the most common anxietydisorders. Furthermore, GAD and PD are the anxiety disorders thathave received considerable attention in clinical practice guidelinesfor general management (primary, secondary and community care)(NICE, 2011) and in specific primary care guidelines (GuidelineWorking Group for the Treatment of Patients with Anxiety

Disorders in Primary Care, 2008), a key setting for the developmentof preventive activities (Starfield et al., 2005; World HealthOrganization, 2008). The risk factor of genes was excluded due toits specificity and complexity; indeed this in itself warrants aseparate review (Gregory et al., 2008). Cohort studies were selectedonly if the effect was not present at baseline, discarding cross-sectional studies given the impossibility of separating cause andeffect in the time. Case-control studies were excluded because theygenerally involve more bias than cohort studies (Rothman et al.,2013). Finally, we focused the review on the community setting inorder to prevent a population bias towards the more severe cases ofanxiety and thus extend the external validity of the review.

2.3. Summary measures

The summary measures included in the studies selected wererelative risk, odds ratio, incidence risk ratio and hazard ratio of aparticular risk factor for the incidence of PD or GAD.

2.4. Data synthesis

We developed a data extraction sheet, pilot tested it on fourrandomly-selected included studies, and refined it accordingly.The main characteristics of these studies were rigorously extractedby PMP, and verified by a second researcher (SCC). Any discre-pancies were resolved by discussion between the two researchers.In the event of disagreement a third researcher (AF) decided.

For each study information was collected about year of pub-lication, author(s), study country, follow-up time of the cohort,sample size, age of the participants at baseline, outcome variable(GAD and/or PD), outcome measure, associated risk factor(s),quantification of the risk associated with exposure, non-associated risk factor(s), and risk of bias (Table 2).

2.5. Assessment of risk of bias

The risk of bias in the studies was assessed independently byPMP and SCC using The Newcastle- Ottawa Scale (NOS) for cohortstudies (Wells et al., 2011). In the event of disagreement, a thirdresearcher (EM) was consulted. The inter-researcher agreementbetween PMP and SCC measured with the Kappa statistic was 0.72(95% CI: 0.56–0.88). The NOS awards stars for three categories:“Selection”, “Comparability”, and “Outcome”, each divided intofurther subcategories. Each study can be awarded a maximum ofone star for each subcategory whilst “Comparability” can beawarded a maximum of two stars (Table 3). The maximumnumber of stars that can be achieved in a study is 9, whichindicates complete absence of bias.

3. Results

3.1. Search results

The search strategy produced 845 potentially relevant articles(370 in Embase, 242 in PubMed and 233 in PsycINFO). A further 20studies were identified from among the references of the articlesthat were eventually selected. Of these 865 articles, 168 hadduplicate titles and were therefore discarded. Of the 697 remain-ing, 534 were excluded after reviewing the title and the abstract,for the following reasons: 289 did not evaluate the incidence of PDor GAD, 52 were not original studies, 46 were undertaken inclinical samples, 43 identified genetic risk factors, 41 were cross-sectional, 34 were case-control studies, and 29 concerned personsyounger than 18 years of age at the end of the study. Afterreviewing the full text of the remaining articles, another 142 were

P. Moreno-Peral et al. / Journal of Affective Disorders 168 (2014) 337–348 339

Table 2Details of studies included in this systematic review.

Countrya

studyCohortfollow-up

Sample(N)

Age atbaseline(years)

Cluster risk factor Outcomevariableb

Outcomemeasurec

Significant associated risk factord Non-significant associated risk factore Assessmentof risk ofbias (NOS)f

US:Johnsonet al.(2000)

6 years 688 14–22 Smoking PD andGAD

DISC ORn; OR; 9PD: Cigarette smoking in adolescence: 15.58 PD and GAD: NoneGAD: Cigarette smoking in adolescence: 5.53nAdjusted for sociodemographic,psychosocial and physical and mental healthvariables

US: Grantet al.(2009)

3 years 34,653 Z18 Age; Female; Educational level;Race-ethnicity; Marital status;Socioeconomic level /financialresources; Alcohol problems; Otheranxiety disorders; Other mentalhealth problems

PD andGAD

AUDADIS-IV ORn; OR; 9PD: Age 20–29: 2.7; Age 30–54: 2.6; Men: 0.5;Family income 0–19,999$: 2.3; GAD: 2.6;Posttraumatic stress disorder: 2.6; Bipolar I:2.3; Personality disorder schizotypal: 2.6;Borderline: 2.3

PD: Education less than high school: 1.5;Hispanic: 0.9; Separated/divorced/widowed:0.9; Alcohol dependence: 1.3; Alcohol abuse:1.0; Any drug dependence: 1.0; Any drugabuse: 1.1; Social phobia: 1.3; Specificphobia: 1.0; Major depression: 1.7; Bipolar II:2.1; Dysthymia: 1.7; ADHD: 1.5; Personalitydisorder obsessive-compulsive: 1.1;Paranoid: 1.0; Schizoid: 0.8; Antisocial: 0.5;Histrionic: 0.7; Narcissistic: 1.0; Avoidant:0.3; Dependent: 1.0

GAD: Age 30–54: 2.9; Men: 0.5; Hispanic:0.5; Separated/divorced/widowed: 1.6;Family income 0–19,999$: 2.1; PD: 1.7; Socialphobia: 1.8; Major depression: 1.9; Bipolar I:2.6; Personality disorder schizotypal: 2.6;Borderline: 4.7; Narcissistic: 2.3nAdjusted for sociodemographic and mentalhealth variables

GAD: High school education: 0.6; Alcoholdependence/abuse: 1.2; Any drugdependence/abuse: 0.9; Specific phobia: 1.1;Posttraumatic stress disorder: 1.5; Bipolar II:1.6; Dysthymia: 1.6; ADHD: 1.9; Personalitydisorder obsessive-compulsive: 0.8;Paranoid: 0.9; Schizoid: 0.7; Antisocial: 0.8;Avoidant: 0.9; Dependent: 0.4

US: Chouet al.(2011)

3 years 13,489 Z50 Alcohol problems PD AUDADIS-IV ORn; OR; 9PD: Occasional binge drinking in femalesubjects: 2.23

PD: None

nAdjusted for sociodemographic, psychosocialand physical and mental health variables

US: Kinleyet al.(2011)

3 years 34,653 Z18 Other anxiety disorders PD andGAD

AUDADIS-IV ORn; OR; 9PD: Lifetime panic attack: 2.73 PD and GAD: NoneGAD: Lifetime panic attack: 1.70nAdjusted for sociodemographic and mentalhealth variables

GE:Knappeet al.(2012)

10 years 3021 14–24 Parental history of mental disorders PD DIA-X/M-CIDI

ORn; OR; 9PD: Specific phobias parents (any): 2.07; PDand agoraphobia parents: 3.38

PD: Agoraphobia without PD parents (any):1.61; Social phobia parents (any): 0.97; GADparents (any): 1.60n Adjusted for sociodemographic variables

PD without agoraphobia parents (any): 3.53;Panic attack parents (any): 3.93nAdjusted for sociodemographic andphysical and mental health variables

AU:Batter-hamet al.(2012)

4 years 3636 20–24and 40–44

Other mental health problems PD andGAD

PHQ-9(DSM-IV)

ORn; OR; 9PD: Sleep disturbance: 1.46 PD and GAD: NoneGAD: Sleep disturbance: 1.22nAdjusted for sociodemographic,psychosocial and physical and mental healthvariables

GE: Liebet al.(2002)

4 years 2427 14–24 Parental history of mental disorders PD andGAD

M-CIDI/DSM-IV

ORn; OR; 8PD: Major depression in both parents: 2.9 PD: Major depression in one parent: 1.6

GAD: None

P.Moreno-Peral

etal./

JournalofAffective

Disorders

168(2014)

337–348

340

GAD: Major depression in one parent: 2.5;Major depression in both parents: 3.7nAdjusted for sociodemographic variables

GE:Isenseeet al.(2003)

4 years 2548 14–24 Smoking PD andGAD

DIA-X/M-CIDI

ORn; OR; 8PD: Dependent regular smokers: 3.3 PD: NoneGAD: Nondependent regular smokers: 0.1 GAD: NonenAdjusted for sociodemographic variables

US:Kendleret al.(2003)

5 years 7322 Z13 Stressful life events in the previousmonth

GAD DSM-III-R(mentalhealthprofessional)

Hazard-ratiosn; Hazard-ratios; 8GAD: Loss: 1.35; Danger: 1.18 GAD: Humiliation: 1.01; Entrapment: 0.98nAdjusted for sociodemographic and mentalhealth variables

NL:Scho-everset al.(2005)

3 years 1915 65–84 Age; Female; Educational level;Marital status; Social support;Chronic diseases; Limitations/Disabilities; History of psychologicalproblems

GAD GMS-AGECAT

ORn; OR; 8GAD: Personal history of depression/anxiety:2.58

GAD: Age474: 1.33; Female: 2.62; Onlybasic education: 1.16; Not/no longermarried: 0.56; Social support: 0.53; Chronicillnesses at baseline: 1.37; ADL disability atbaseline: 0.78; IADL disability at baseline:1.31; MMSEo26: 2.02; Recent decrease inADL functioning: 0.60; Recent decrease inIADL functioning: 1.56

nAdjusted for sociodemographic,psychosocial and physical and mental healthvariables

US:Johnsonet al.(2006)

11 years 505 14–22 Other mental health problems PD andGAD

DISC-I ORn; OR; 8PD: Z1 personality disorder: 4.64 PD and GAD: NoneGAD: Z1 personality disorder: 14.41nAdjusted for sociodemographic and mentalhealth variables

NZ: Moffittet al.(2007)

29 years 945 3 Female; Stressful life events duringchildhood; Inhibited temperament;Psychiatric care

GAD DIS (DSM-III-R y DSM-IV)

ORn; OR; 8GAD: Child maltreatment before age 11: 4.49;Received mental health services, age 20–32:6.16; Psychiatric medication, age 20–32: 5.05

GAD: Female: 0.96; Lost a parent before age11: 1.62; Inhibited temperament ages 3–5:2.31

nAdjusted for genderUS: Kessleret al.(2008)

11 years 5001 15–54 Parental history of mental disorders;Stressful life events duringchildhood; Personality

GAD CIDI (DSM-III-R)

ORn; OR; 8GAD: Parental MDE: 1.5; Parental GAD: 1.6;Parental panic disorders: 1.5; Parentalsubstance disorder: 1.5; Childhood adversityneglect: 1.7; Physical abuse: 1.8; Sexual abuse:1.6; Divorce of parents: 1.4; Neuroticism: 1.3

GAD: Death of parent: 1.1; Other long-termseparation in childhood: 1.3; Parental ASPD:1.5; Extroversion: 0.9; Openness toexperience: 1.0

nAdjusted for sociodemographic variablesGE: Beesdoet al.(2010)

10 years 3021 14–24 Stressful life events duringchildhood; Parental characteristics;Personality; Parental history ofmental disorders

GAD DIA-X/M-CIDI

Hazard-ratiosn; Hazard-ratios; 8GAD: Childhood separation events: 2.44;Behavioural inhibition: 1.97; Harm avoidance:1.69; Reward dependence: 1.43; Parental GAD:3.77

GAD: Dysfunctional family functioning: 1.48;Parental emotional warmth: 0.91; Parentalrejection: 1.20; Parental overprotection:1.48; Novelty seeking: 0.94; Resilience: 0.84;Parental anxiety disorder but no depressivedisorder: 1.76; Parental depressive disorderbut no anxiety disorder: 1.77; Parentalanxiety and depressive disorder: 1.46

nAdjusted for sociodemographic and mentalhealth variables

US:Pietrzaket al.(2013)

3 years 34,653 Z18 Other mental health problems PD andGAD

AUDADIS-IV ORn; OR; 8PD: Subsyndromal depression (limitedsymptoms): 1.62; Major depression: 1.75

PD: Subsyndromal depression (no distress):1.45

GAD: Subsyndromal depression (limitedsymptoms): 1.41; Subsyndromal depression(no distress): 2.00; Major depression: 2.41

GAD: None

nAdjusted for sociodemographic variablesSD:Gräsbecket al.(1993)

15 years 2550 o17 Age; Female PD andGAD

DSM-III-TR(mentalhealthprofessional)

IDRn; IDR; 7PD: Age 30–44 years: 1.64; Women: 2.86 PD: NoneGAD: Age 45–64 years: 0.04; Men: 0.00;Women: 0.02

GAD: None

nAdjusted for sociodemographic variables

P.Moreno-Peral

etal./

JournalofAffective

Disorders

168(2014)

337–348

341

Table 2 (continued )US:Kendleret al.(2000)

9 years 1033twinpairs

17–55 Parental characteristics PD andGAD

SCID (DSM-III-R)

ORn; OR; 7Parentsmean:58.61472

parentsPD: no significant GAD: Mother coldness: 1.22;Father coldness: 1.31; Mother protectiveness:1.28; Father protectiveness: 1.41

PD: Mother coldness: 1.21; Fathercoldness:1.30; Mother protectiveness: 1.18;Father protectiveness: 1.08; Motherauthoritarianism:1.26; Fatherauthoritarianism:1.30 GAD: Motherauthoritarianism: 1.16; Fatherauthoritarianism: 1.11

nAdjusted for sociodemographic,psychosocial variables

UK: Clarket al.(2007)

38 years 9297 7 History of psychological problems GAD CIS-R ORn; OR; 7GAD: Internalising at age 7 (case): 1.91;Externalising at age 11 (case): 1.65;Externalising at age 16 (case): 1.61;Internalising at age 16 (borderline): 1.48;Internalising at age 16 (case): 2.01; Cumulativechildhood disorders 1: 1.70; Cumulativechildhood disorders 2: 1.72; Cumulativechildhood disorders Z3: 2.40; Malaise at age33: 4.73

GAD: Externalising at age 7 (case): 1.35;Internalising at age 11 (case): 1.37

nAdjusted for sociodemographic and mentalhealth variables

GE: Rudazet al.(2010)

1 year and5 months

1396 18–24 Number of physical diseasessuffered; Somatic symptoms andbody sensations; Other anxietydisorders; Other mental healthproblems

PD ADIS-IV-L(DSM-IV)

ORn; OR; 7PD: History of physical diseases 0–14: 1.48;Disease phobia: 1.55; Other psychologicaldisorders: 2.47

PD: Somatic symptoms:1.11; Bodysensations: 0.93

nAdjusted for physical and mental healthvariables

SP:Bulbenaet al.(2011)

15 years 158 16–20 Joint hypermobility syndrome PD andGAD

SCID-IV RRn; RR; 7PD: JHS present: 22.3 PD: NoneGAD: None GAD: JHS present: 2.9nNon-adjusted

US: Gelleret al.(2001)

6 months 459 Z18 Miscarriage PD DIS RRn; RR; 4PD: None PD: Miscarriage: 1.8nNon-adjusted

a US: United States; SD: Sweden; NL: Netherlands; NZ: New Zealand; UK: United Kingdom; GE: Germany; AU: Australia; SP: Spain.b PD: Panic disorder; GAD: Generalised anxiety disorder.c DIS: Diagnostic interview schedule; DSM-III-R: Diagnostic and statistical manual of mental disorders third edition revision; DSM-IV: Diagnostic and statistical manual of mental disorders fourth edition; DSM-IV-TR: Diagnostic

and statistical manual of mental disorders fourth edition text revision; AUDADIS-IV: Alcohol use disorder and associated disabilities interview schedule IV; GMS-AGECAT: Geriatric mental state automated geriatric examination forcomputer assisted taxonomy; SCID: Structured clinical interview for DSM disorders; CIDI- SF: Composite international diagnostic interview, short form; ADIS-IV-L: Anxiety disorders interview schedule lifetime version; PHQ-9:Patient health questionnaire; CIS-R: Revised clinical interview schedule; DISC: Diagnostic interview schedule for children. DIA-X/M-CIDI: The Munich-composite international diagnostic interview.

d OR: Odd ratio; IDR: Incidence density ratio (calculated by the researchers, not given in the original article); RR: Relative risk; MDE: Major depression episode; JHS: Joint hypermobility syndrome.e ADHD: Attention-deficit/ hyperactivity disorder; ADL: Activities of daily living; IADL: Instrumental activities of daily living; MMSE: Mini-mental state exam; ASPD: Antisocial personality disorder.f NOS: Newcastle ottawa quality assessment scale cohort studies (higher score"better quality).

P.Moreno-Peral

etal./

JournalofAffective

Disorders

168(2014)

337–348

342

excluded, for the following reasons: 75 did not evaluate theincidence of PD or GAD, 12 identified genetic risk factors, 13 werecross-sectional studies, 11 were not original studies, 10 concernedpersons younger than 18 years of age at the end of the follow-up,8 were carried out in clinical samples, 6 were case-control studies,

6 did not exclude persons with anxiety at baseline, and 1 was aretrospective cohort study (Fig. 1).

Finally, a total of 21 publications were selected; these comprisethe basis of this review. The data from these studies wereextracted and included in Table 2.

Table 3Risk of bias in the studies reviewed.

Study Selection Comparability Outcome

Representativenessof the exposedcohort

Selection ofthe non-exposedcohort

Ascertainmentof exposure

Outcome wasnot present atstart of study

Study controls forgender/study controls forany additional factor

Assessmentof outcome

Was follow-up longenough foroutcomes to occura

Adequacyof follow-up ofcohorts

Batterhamet al.(2012)

n n n n n/n n n n

Beesdoet al.(2010)

n n n n/n n n n

Bulbenaet al.(2011)

n n n n n n n

Clark et al.(2007)

n n n n/n n n

Chou et al.(2011)

n n n n n/n n n n

Gelleret al.(2001)

n n n n

Grant et al.(2009)

n n n n n/n n n n

Gräsbecket al.(1993)

n n n n n/ n n

Isenseeet al.(2003)

n n n n n/n n n

Johnsonet al.(2000)

n n n n n/n n n n

Johnsonet al.(2006)

n n n n n/n n n n

Kendleret al.(2003)

n n n n n/n n n

Kendleret al.(2000)

n n n n/n n n

Kessleret al.(2008)

n n n n n/n n n

Kinleyet al.(2011)

n n n n n/n n n n

Knappeet al.(2012)

n n n n n/n n n n

Moffittet al.(2007)

n n n n n/ n n n

Pietrzaket al.(2013)

n n n n/n n n n

Rudazet al.(2010)

n n n n /n n n

Schoeverset al.(2005)

n n n n n/n n n

Lieb et al.(2002)

n n n n n/n n n

a Follow-up was considered to be adequate with effect from 6 months, based on the current definition of generalised anxiety disorder.

P. Moreno-Peral et al. / Journal of Affective Disorders 168 (2014) 337–348 343

The studies were based on community samples in USA (N"10),Germany (N"5), The Netherlands (N"1), New Zealand (N"1),Australia (N"1), United Kingdom (N"1), Sweden (N"1) andSpain (N"1). Of the 21 studies, 20 (95%) had been published since2000. The follow-up of the cohorts varied from 6 months to38 years (median"5 years). Overall, the studies evaluated a totalof 163,366 persons. The number of participants in each studyvaried from 158 to 34,653 (median"2550) and all persons wereaged 14 years or older at baseline. For the dependent variablestudied, 6 (28.5%) studies referred to GAD, 4 (19.1%) to PD, and theother 11 (52.4%) to both PD and GAD. In the 21 studies included inour systematic review a total of 81 different risk factors potentiallyassociated with the onset of GAD and/or PD were evaluated. Theserisk factors were involved in 162 different evaluations of associa-tion, of which 85 were not significant; among the latter, only 5 hadOR–RR greater than 2 (see Table 2), with a potential problemarising in several cases of lack of statistical power; such is the case,for example, of the association between joint hypermobilitysyndrome and GAD (Bulbena et al., 2011).

3.2. Risk of bias in individual studies

Table 3 presents data on the risk of bias in each study. The totalmean NOS was 7.9 (SD"1.15; range 4–9). The 21 cohort studies wererepresentative of their community. In all studies except one theexposed cohort belonged to the same setting as the non-exposedcohort; in 17 of the 21 studies the risk factors were evaluatedobjectively. All the studies demonstrated that outcome of interestwas not present at the start of the study; 18 of the 21 controlled in thestudy design or analysis for gender, and 17 of the 21 controlled for anyother factor. In all 21 studies the outcome was assessed by mentalhealthcare professionals or diagnostic structured interviews, and in all

studies except one the follow-up period was considered adequate.More than 6 months was established as an adequate follow-up periodas the current definition of GAD implies the presence of symptoms forat least six months (APA, 2013). Finally, the losses during the follow-upperiod did not exceed 15% in 11 of the 21 studies, in the remainingstudies drop-outs were between 16% and 55.4% (three articles did notprovide drop-out figures).

In order to improve the accessibility of the findings on riskfactor, these factors were divided into three main categories:sociodemographic, psychosocial, and mental and physical health.

3.3. Sociodemographic risk factors

3.3.1. Panic disorderConcerning age associated with the onset of PD, Grant et al.

(2009) reported an age of 20–54 years and Gräsbeck et al. (1993)reported an age of 30–44 years. Concerning gender, both of thetwo articles that assessed this variable found a significant associa-tion between female gender and onset of PD (Grant et al., 2009).The educational level was not associated with the onset of PD ineither of the studies where it was evaluated (Grant et al., 2009;Schoevers et al., 2005). Nor was being Black or Hispanic, ascompared with White, associated with the onset of PD in the onlystudy where this was assessed (Grant et al., 2009). In addition toother factors, the study by Grant et al. (2009) also examinedmarital status, which was not significant, and a low socioeconomicstatus, which was significant.

3.3.2. Generalised anxiety disorderConcerning age associated with the onset of GAD, Grant et al.

(2009) found an age of 30–54 years and Gräsbeck et al. (1993) an

Fig. 1. Flow chart of the studies reviewed and finally included in the review.

P. Moreno-Peral et al. / Journal of Affective Disorders 168 (2014) 337–348344

age of 45–64 years. Schoevers et al. (2005) however, in personsaged 65 years and over, found no relation between age and theonset of GAD. Concerning gender, of the four studies where thiswas assessed, two found a significant association between beingfemale and the risk of GAD (Grant et al., 2009; Gräsbeck et al.,1993), whereas the other two failed to find this association(Moffitt et al., 2007; Schoevers et al., 2005). A low educationallevel was not associated with the onset of GAD in either of the twostudies that assessed this variable (Grant et al., 2009; Schoeverset al., 2005). Grant et al. (2009) noted that Hispanics, in compar-ison with Blacks and non-Hispanic individuals, had a lower risk ofGAD. Marital status was assessed in two studies: one found thatbeing divorced or widowed was related with GAD (Grant et al.,2009) but the other did not (Schoevers et al., 2005). Finally, theone study that examined the relation between economic resourcesand the onset of GAD found that having few economic resourcesincreased the likelihood of the onset of GAD (Grant et al., 2009).

The sociodemographic risk factors, classified according to theirassociation with the onset of PD or GAD, are shown in Table 4.

3.4. Psychosocial risk factors

3.4.1. Panic disorderSmoking during adolescence (Johnson et al., 2000) and being a

dependent regular smoker (Isensee et al., 2003) were associatedwith the onset of PD in adulthood. Alcohol problems (bingedrinkers) were associated with the onset of PD in the only studythat examined it (Chou et al., 2011). The parental characteristicswere assessed in one study, but no association was found betweenthe onset of PD and cold, protective or authoritarian parents(Kendler et al., 2000).

3.4.2. Generalised anxiety disorderThe onset of GAD in adulthood was related with smoking during

adolescence (Johnson et al., 2000), though contradictory findingsabout smoking and the onset of GAD were also found (Isensee et al.,2003). Three studies examined stressful life events during childhood inrelation to the onset of GAD, with all three finding significantassociations (Beesdo et al., 2010; Kessler et al., 2008; Moffitt et al.,2007). One study looked at the relation between the onset of GAD andstressful life events during the previous month, finding a positiveassociation (Kendler et al., 2003). The characteristics of the parentswere examined in two studies. Kendler et al. (2000) found that havingcold and protective parents increased the risk of GAD in their children,whereas Beesdo et al. (2010), however, found no significant associa-tions between the parental characteristics (parental emotionalwarmth, parental rejection and parental overprotection) and GAD.

One study evaluated an inhibited temperament during child-hood but found no association between this and the onset of GAD(Moffitt et al., 2007). Personality traits, such as neuroticism(Kessler et al., 2008), behavioural inhibition, harm avoidance andreward dependence (Beesdo et al., 2010), at any stage of life, werefound to be significantly associated with GAD. And finally, con-cerning social support, Schoevers et al. (2005) found no significantassociations in a study of persons aged 65 years and over.

The psychosocial risk factors, classified according to theirassociation with the onset of PD or GAD, are shown in Table 4.

3.5. Physical and mental health risk factors

3.5.1. Panic disorderOne of the studies examining physical health problems

found no significant association between miscarriage and PD

Table 4Associated, non-associated and controversial risk factors for the onset of generalised anxiety disorder and panic disorder.

Panic disorder Generalised anxiety disorder

Sociodemographicfactors

Psychosocialfactors

Physical andmental healthfactors

Sociodemographicfactors

Psychosocial factors Physical andmental healthfactors

Associatedriskfactorsn

Age 20–54 years (1/1) Smoking (2/2) Nº of physicaldiseases suffered (1/1)

Age 30–54 years (1/1) Stressful life events during childhood (3/3) Parental historyof mentaldisorders (3/3)

Age 30–44 years (1/1) Alcoholproblems (1/1)

Joint hypermobilitysyndrome (1/1)

Age 45–64 (1/1) Stressful life events during the previousmonth (1/1)

Other anxietydisorders (3/3)

Female gender (2/2) Parental history ofmental disorders (2/2)

Race-ethnicity non-Hispanic and Black (1/1)

Personality: behavioural inhibition, harmavoidance, reward dependence andneuroticism (2/2)

Other mentalhealth problems(4/4)

Low socioeconomiclevel / financialresources (1/1)

Other anxietydisorders (3/3)

Marital statusseparated /divorced/widowed (1/1)

History ofpsychologicalproblems (2/2)

Other mental healthproblems (5/5)

Low socioeconomiclevel/ financialresources (1/1)

Psychiatric care(1/1)

Non-associatedriskfactorsnn

Educational level (0/2) Parentalcharacteristics(0/1)

Miscarriage (0/1) Age Z65 years (0/1) Low social support (0/1) Jointhypermobilitysyndrome (0/1)

Race-ethnicityHispanic and Black (0/1)

Somatic symptomsand body sensations(0/1)

Low educational level(0/2)

Inhibited temperament in childhood (0/1) Any chronicdisease (0/1)

Marital status (0/1) Limitations/disabilities(0/1)

Controversialriskfactorsnnn

Female gender (2/4) Smoking (1/2)Parental characteristics (1/2)

(/) Numerator"number of articles finding a significant association (po0.05); denominator"number of articles analysing the risk factor of interest.n Number of studies finding a significant association (po0.05) above the median.nn Number of studies finding a significant association (po0.05) below the median.nnn Number of studies finding a significant association (po0.05) equal to the median.

P. Moreno-Peral et al. / Journal of Affective Disorders 168 (2014) 337–348 345

(Geller et al., 2001). Rudaz et al. (2010) found a significantassociation between the number of physical disorders experiencedfrom a list of 14 (e.g., heart problems, asthma, migraine) and thefuture development of PD. In addition, these authors also foundthat somatic symptoms (evaluated with the Whiteley Index) andbody sensations (evaluated with the Body Sensations Question-naire) were not associated with PD. Finally, another study foundthat the joint hypermobility syndrome was significantly associatedwith developing PD (Bulbena et al., 2011).

A parental history of depression (Lieb et al., 2002) or anxiety(Knappe et al., 2012) was associated with the onset of PD in theirchildren. The presence of other anxiety disorders in the patientwas found to be related with PD in three studies (Grant et al.,2009; Kinley et al., 2011; Rudaz et al., 2010). The presence of othermental health problems in the patient (other than anxiety) wasassociated with PD in 5 studies. These problems concernedpersonality disorders (Grant et al., 2009; Johnson et al., 2006),bipolar disorder (Grant et al., 2009), sleep disturbance (Batterhamet al., 2012), depression (Pietrzak et al., 2013), and other psycho-logical disorders (Rudaz et al., 2010).

3.5.2. Generalised anxiety disorderConcerning physical health problems, the joint hypermobility

syndrome was not significantly associated with the onset of GADin one study, although its OR was 2.9 (Bulbena et al., 2011).Another study involving individuals aged 65 years and over foundthat suffering any chronic disease was not predictive of GAD(Schoevers et al., 2005). In this same population, Schoevers et al.(2005) also examined the limitations/incapacities associated withthese disorders, but again found no significant association with theonset of GAD.

A parental history of GAD (Beesdo et al., 2010), depression (Liebet al., 2002), or other mental health problems (Kessler et al., 2008)were associated with GAD in their children. The presence of otheranxiety disorders in the patient was found to be related with theonset of GAD in three studies (Grant et al., 2009; Kinley et al.,2011; Rudaz et al., 2010). The presence of other mental healthproblems in the patient was also associated with a greater like-lihood of developing GAD in 4 studies. These problems werepersonality disorders (Grant et al., 2009; Johnson et al., 2006),bipolar disorder, major depression (Grant et al., 2009), sleepdisturbance (Batterham et al., 2012), and depression (Pietrzaket al., 2013).

Having suffered anxiety or depression in the past was a riskfactor for GAD in one study of individuals aged 65 years and over(Schoevers et al., 2005). Another study (Clark et al., 2007) foundthat internalising (worried, solitary, miserable, fearful, and fussy)and externalising disorders (destructive, fights, not much liked byother children, irritable, disobedient, lies, steals, resentful/aggres-sive and bullies) in childhood or adolescence, the accumulation ofthese disorders and malaise (psychological and somatic symp-toms) increased the risk for GAD. Finally, Moffitt et al. (2007)found that being in receipt of psychiatric care for any mentalhealth problem (other than GAD) also predicted its appearance.

The physical and mental health risk factors, classified accordingto their association with the onset of PD or GAD, are shown inTable 4.

4. Discussion

4.1. Main results

A group of sociodemographic and psychosocial factors accom-panied by mental and physical health conditions contribute to anincreased risk for the onset of PD and GAD. Our study is the first, to

our knowledge, to provide a comprehensive overview of riskfactors (excluding genetic risk factors) for the onset of PD andGAD based on cohort studies in general adult populations. Thesefindings may prove useful for the development of preventiveinterventions.

4.2. Strengths and limitations

This systematic review involved a large number of risk factorsfor PD and GAD, generating evidence about the importance of eachof these factors from a global and multicausal approach. Thequality of the methodology of the studies included in thissystematic review was excellent in all cases (absence of risk ofbias as per the NOS criteria). In addition, as we only includedcohort studies we minimised the possibility of bias relating tomeasuring the risk factors and the effect, at the same timeguaranteeing that any possible cause preceded the effect. Mostof the studies finally included provided adjusted results, at leastfor sociodemographic variables, resulting in robust OR and RRfigures for each risk factor. Finally, the findings of our systematicreview have wide external validity as they all involved communitystudies.

Concerning the limitations, in general we only found a fewstudies for each of the risk factors identified using our inclusionand exclusion criteria. This highlights the need for more prospec-tive cohort studies in the general population. The search filteretiology/narrow used in this review is quite specific (95%) but notso sensitive (51%) (Haynes et al., 1994; Wong et al., 2003). Thismeans that some relevant studies may not have been included inthe review. In addition, by excluding case-control studies we mayhave missed some high quality analytical studies with little bias, asfor example the study by Keyl and Eaton (1990). Although we haveminimised the heterogeneity of the studies included by means ofour inclusion and exclusion criteria (cohort studies, generalpopulation, anxiety incidence as main outcome), a certain degreeof heterogeneity still remains due to variations in follow-up time,sample size and age ranges.

4.3. Comparison with other systematic reviews

4.3.1. Sociodemographic factorsA systematic review in 2004 identified female gender, middle

age, and financial difficulty as factors associated with anxietydisorders in Pakistan (Mirza and Jenkins, 2004). Another review,involving children, adolescents and adults (but evaluating riskfactors in childhood) also found that female gender and financialdifficulty in the family were risk factors for anxiety disorders(Beesdo et al., 2009). Female gender was also identified as a riskfactor for anxiety disorders in the elderly (Vink et al., 2008). In ourreview, though, female gender was only found to be clearlyassociated with PD but was doubtful for GAD, due to the lack ofassociation in the study by Schoevers et al. (2005), undertaken inindividuals aged 65 years or over. However, if we exclude thislatter study, the proportion of studies giving significant results forgender in GAD would be 2/3 instead of 2/4 (Table 4).

A low educational level and being separated, divorced orwidowed were not found to be associated with the onset of PDor GAD in our review. The review by Mirza and Jenkins (2004) inPakistan, however, did find these factors to be associated with theonset of anxiety disorders.

When we compared the sociodemographic factors identified inour review with other reviews conducted in children and adoles-cents, we found that the association between female gender andthe onset of anxiety disorders in children and adolescents wasweak, but increased with age (Beesdo et al., 2009; Beesdo-Baumand Knappe, 2012). Non-Hispanic and black race were associated

P. Moreno-Peral et al. / Journal of Affective Disorders 168 (2014) 337–348346

with the onset of GAD and/or PD in both children-adolescents andadults. Divorce, low economic status and lower educational levelwere also associated in both adults and children-adolescents,although for the last three it referred to the situation of theirparents (Beesdo et al., 2009; Beesdo-Baum and Knappe, 2012).

4.3.2. Psychosocial factorsThe lack of empirical support received by social network in

relation to the onset of GAD is also reflected in the reviewundertaken by Vink et al. (2008), though this study was under-taken in the elderly and assessed anxiety in general.

Our review found no cohort study examining the associationbetween stressful life events and the onset of PD. However, thereview by Klauke et al. (2010) of stressful life events and thegenesis of PD found an association with the factors threat (dangerof death, either personal or of a cohabiting relative), health relatedevents, emotional/physical/sexual abuse experiences in childhood,and interpersonal life events of loss or separation. This differenceis explained by the fact that the studies included in the review byKlauke et al. (2010) were case-control studies and the cohortstudies referred to panic attack instead of PD.

In our review we found that stressful life events in childhoodand adulthood predicted the onset of GAD in adult population.Stressful life events in childhood also predicted the onset ofanxiety disorders in children and adolescents (Beesdo et al.,2009). Parental characteristics (harsh discipline, autonomy-granting and overinvolvement) and behavioural inhibition werealso identified as risk factors in children and adolescents (Beesdoet al., 2009). In the adult population, though, only behaviouralinhibition was a significant factor.

As found in our review, smoking during adolescence wasassociated with the onset of PD in adulthood in the review byCosci et al. (2010).

4.3.3. Physical and mental health factorsPhysical health problems were also found to be associated with

the onset of anxiety disorders in other reviews (Gandy et al., 2012;Klauke et al., 2010). Klauke et al. (2010) found that stressful lifeevents related with physical health in the affected person wereassociated with PD. Gandy et al. (2012), on the other hand, in asystematic review of the psychosocial predictors of anxiety inadults with epilepsy, found that “illness representations” evalu-ated by the Illness Perception Questionnaire was associated withanxiety disorders. In our review the number of physical diseasesexperienced (Rudaz et al., 2010) and the joint hypermobilitysyndrome were associated with the onset of PD (Bulbena et al.,2011), but miscarriages and somatic symptoms and body sensa-tions were not (Geller et al., 2001; Rudaz et al., 2010). However, nophysical health problem was associated with the development ofGAD (Schoevers et al., 2005).

As found in our review, mental health problems were found tobe associated with anxiety disorders in different reviews (Gandyet al., 2012; Vink et al., 2008). Vink et al. (2008) reviewed riskfactors associated with anxiety in the elderly and found thatpsychopathology was associated with anxiety disorders. Gandyet al. (2012), though, found that depression was associated withanxiety disorders in adults with epilepsy.

As in our review, a parental history of mental disorders wasalso found to be associated with anxiety disorders in their childrenin the review by Beesdo et al. (2009). This was the only physical ormental health risk factor associated with anxiety disorders in bothadults and in children-adolescents (Beesdo et al. 2009; Beesdo-Baum and Knappe, 2012).

4.4. Practical implications

The joint approach examining different factors identified in theliterature as associated with PD or GAD enabled us to obtain amore global understanding of the risk for these disorders. Ourreview argues in favour of selecting those risk factors that need tobe considered in order to obtain risk algorithms predicting theonset of PD and GAD. Currently, there exists a risk algorithm topredict the onset of anxiety disorders in primary care (King et al.,2011). Based on a selection of risk factors, information can becalculated about the level (risk likelihood) and profile (co-occur-rence of different risk factors for each patient) of risk. This could beuseful when developing and implementing individualised anxietyprevention programs for each patient (Bellón et al., 2013).

Given the small number of cohort studies for each risk factoridentified in our review, further research is needed contemplatinga joint approach of these risk factors followed over a long period oftime and modelled on large population-based samples.

Some of the risk factors identified in this review as precursorsof the onset of PD and GAD can be modified by socio-politicalinterventions (e.g., low educational level, low level of income,smoking) or healthcare interventions (e.g., physical or mentalhealth problems, addictions to smoking and alcohol). Furtherstudies are needed to establish whether manipulation of thesefactors may actually reduce the onset of PD and/or GAD. Risk factorresearch could be used for prevention programs and reciprocally,such prevention knowledge could be used for risk factor research(Zvolensky et al., 2006).

5. Conclusion

The results of this systematic review show the risk factorscontributing to the onset of PD and GAD in the general adultpopulation. These data should prove useful for the development ofinterventions based on a conception of global risk for the preven-tion of PD and GAD.

Conflict of interestThe authors all declare they have no competing interests.

Role of funding sourceNone.

AcknowledgementsThe authors thank the Primary Care District of Malaga, particularly Maximi-

liano Vilaseca, for their support. We also thank the Institute of Health Carlos III(ISCIII) (RD12/0005), the Andalusian Public Foundation for Health and BiomedicineResearch in Malaga (FIMABIS) and Boni Bolíbar and the Spanish Network ofPrimary Care Research (redIAPP) for their economic and logistical support.

References

Alonso, J., Angermeyer, M.C., Bernert, S., Bruffaerts, R., Brugha, T.S., Bryson, H., deGirolamo, G., Graaf, R., Demyttenaere, K., Gasquet, I., Haro, J.M., Katz, S.J.,Kessler, R.C., Kovess, V., Lèpine, J.P., Ormel, J., Polidori, G., Russo, L.J., Vilagut, G.,Almansa, J., Arbabzadeh-Bouchez, S., Autonell, J., Bernal, M., Buist-Bouwman, M.A.,Codony, M., Domingo-Salvany, A., Ferrer, M., Joo, S.S., Martínez-Alonso, M.,Matschinger, H., Mazzi, F., Morgan, Z., Morosini, P., Palacín, C., Romera, B., Taub, N.,Vollebergh, W.A., 2004. ESEMeD/MHEDEA, 2004. Prevalence of mental disorders inEurope: results from the European Study of the Epidemiology of Mental Disorders(ESEMeD) project. Acta Psychiatr. Scand. Suppl. 420, 21–27.

American Psychiatric Association, 2013. Diagnostic and Statistical Manual of MentalDisorders, 5th ed. American Psychiatric Publishing, Arlington, Virginia.

Andlin-Sobocki, P., Wittchen, H.-U., 2005. Cost of anxiety in Europe. Eur. J. Neurol.12, 39–44.

Batterham, P., Glozier, N., Christensen, H., 2012. Sleep disturbance, personality andthe onset of depression and anxiety: prospective cohort study. Aust. N. Z. J.Psychiatry 46 (11), 1089–1098.

P. Moreno-Peral et al. / Journal of Affective Disorders 168 (2014) 337–348 347

Baxter, A.J., Vos, T., Scott, K.M., Ferrari, A.J., Whiteford, H.A., 2014. The global burdenof anxiety disorders in 2010. Psychol. Med. 22, 1–12.

Beesdo, K., Knappe, S., Pine, D.S., 2009. Anxiety and anxiety disorders in childrenand adolescents: developmental issues and implications for DSM-V. Psychiatr.Clin. North Am. 32, 483–524.

Beesdo, K., Pine, D.S., Lieb, R., Wittchen, H.U., 2010. Incidence and risk patterns ofanxiety and depressive disorders and categorization of generalized anxietydisorder. Arch. Gen. Psychiatry 67 (1), 47–57.

Beesdo-Baum, K., Knappe, S., 2012. Developmental epidemiology of anxietydisorders. Child Adolesc. Psychiatr. Clin. North Am. 21, 457–478.

Bellón, J.A., Conejo-Cerón, S., Moreno-Peral, P., King, M., Nazareth, I., Martín-Pérez, C.,Fernández-Alonso, C., Ballesta-Rodríguez, M.I., Fernández, A., Aiarzaguena, J.M.,Montón-Franco, C., Ibañez-Casas, I., Rodríguez-Sánchez, E., Rodríguez-Bayón, A.,Serrano-Blanco, A., Gómez, M.C., Lafuente, P., Del Mar Muñoz-García, M.,Mínguez-Gonzalo, P., Araujo, L., Palao, D., Espinosa-Cifuentes, M., Zubiaga, F.,Navas-Campaña, D., Mendive, J., Aranda-Regules, J.M., Rodríguez-Morejón, A.,Salvador-Carulla, L., de Dios Luna, J., 2013. Preventing the onset of majordepression based on the level and profile of risk of primary care attendees:protocol of a cluster randomised trial (the predictD-CCRT study). BMC Psychiatry13, 171.

Bulbena, A., Gago, J., Pailhez, G., Sperry, L., Fullana, M.A., Vilarroya, O., 2011. Jointhypermobility syndrome is a risk factor trait for anxiety disorders: a 15-yearfollow-up cohorte study. Gen. Hosp. Psychiatry 33, 363–370.

Chou, K.L., Mphil, L.K., Mackenzie, C.S., 2011. Binge drinking and Axis I psychiatricdisorders in community-dwelling middle-aged and older adults: results fromthe National Epidemiologic Survey on Alcohol and Related Conditions(NESARC). J. Clin. Psychiatry 72, 640–647.

Clark, C., Rodgers, B., Cadwell, T., Power, C., Stansfeld, S., 2007. Childhood andadulthood psychological ill health as predictors of midlife affective and anxietydisorders. Arch. Gen. Psychiatry 64, 668–678.

Cosci, F., Knuts, I.J., Abrams, K., Griez, E.J., Schruers, K.R., 2010. Cigarette smokingand panic: a critical review of the literature. J. Clin. Psychiatry 71, 606–615.

Davydow, D.S., Katon, W.J., Zatzick, D.F., 2009. Psychiatric morbidity and functionalimpairments in survivors of burns, traumatic injuries, and ICU stays for othercritical illnesses: a review of the literature. Int. Rev. Psychiatry 21, 531–538.

Davydow, D.S., Richardson, L.P., Zatzick, D.F., Katon, W.J., 2010. Psychiatric morbid-ity in pediatric critical illness survivors: a comprehensive review of theliterature. Arch. Pediatr. Adolesc. Med. 164, 377–385.

Feldner, M.T., Zvolensky, M.J., Schmidt, N.B., 2004. Prevention of anxiety psycho-pathology: a critical review of the empirical literature. Clin. Psychol.: Sci. Pract.11, 405–424.

Fleiss, J.L., 1981. Statistical Methods for Rates and Proportions, 2nd ed. John Wiley,New York.

Gandy, M., Sharpe, L., Perry, K.N., 2012. Psychosocial predictors of depression andanxiety in patients with epilepsy: a systematic review. J. Affect. Disord. 140,222–232.

Gariepy, G., Nitka, D., Schmitz, N., 2010. The association between obesity andanxiety disorders in the population: a systematic review and meta-analysis. Int.J. Obes. (Lond.) 34, 407–419.

Geller, P., Klier, C.M., Neugebauer, R., 2001. Anxiety disorders following miscarriage.J. Clin. Psychiatry 62, 432–438.

Gili, M., Comas, A., Garcia-Garcia, M., Monzon, S., Serrano-Blanco, A., Roca, M., 2010.Comorbidity between common mental disorders and chronic somatic diseasesin primary care patients. Gen. Hosp. Psychiatry 32, 240–245.

Grant, B.F., Goldstein, R.B., Chou, S.P., Huang, B.M.D., Stinson, F.S., Dawson, D.A.,Saha, T.D., Smith, S.M., Pulay, A.J., Pickering, R.P., Ruan, W.J., Compton, W.M.,2009. Sociodemographic and psychopathologic predictors of first incidence ofDSM-IV substance use, mood, and anxiety disorders: results from the wave2 national epidemiologic survey on alcohol and related conditions. Mol.Psychiatry 14, 1051–1066.

Gräsbeck, A., Hagnell, O., Otterbeck, L., Rorsman, B., 1993. Anxiety in the Lundbystudy: re-evaluation according to DSM-II-R, incidence and risk. Neuropsycho-biology 27, 1–8.

Gregory, A.M., Lau, J.Y., Eley, T.C., 2008. Finding gene-environment interactions forgeneralised anxiety disorder. Eur. Arch. Psychiatry Clin. Neurosci. 258, 69–75.

Guideline Working Group for the Treatment of Patients with Anxiety Disorders inPrimary Care, 2008. Madrid: National Plan for the NHS of the MSC. HealthTechnology Assessment Unit. Laín Entralgo Agency. Community of Madrid.

Haynes, R.B., Wilczynski, N., McKibbon, K.A., Walker, C.J., Sinclair, J.C., 1994.Developing optimal search strategies for detecting clinically sound studies inMEDLINE. J. Am. Med. Inform. Assoc. 1, 447–458.

Isensee, B., Wittchen, H.U., Stein, M.B., Höfler, M., Lieb, R., 2003. Smoking increasesthe risk of panic. Arch. Gen. Psychiatry 60, 692–700.

Johnson, J.G., Cohen, P., Kasen, S., Brook, J.S., 2006. Personality disorders evident byearly adulthood and risk for anxiety disorders during middle adulthood. J.Anxiety Disord. 20, 408–426.

Johnson, J.G., Cohen, P., Pine, D.S., Klein, D.F., Kasen, S., Brook, J.S., 2000. Associationbetween cigarette smoking and anxiety disorders during adolescence and earlyadulthood. JAMA 284, 2348–2351.

Kendler, K.S., Hettema, J.M., Butera, F., Gadner, C.O., Prescott, C.A., 2003. Life eventdimensions of loss, humiliation, entrapment and danger in the prediction ofonsets of major depression and generalized anxiety. Arch. Gen. Psychiatry 60,789–796.

Kendler, K.S., Myers, J., Prescott, C.A., 2000. Parenting and adult mood, anxiety andsubstance use disorders in female twins: an epidemiological, multi-informant,retrospective study. Psychol. Med. 30, 281–294.

Kessler, R.C., Chiu, W.T., Demler, O., Walters, E.E., 2005. Prevalence, severity, andcomorbidity of 12-month DSM-IV disorders in the National Comorbidity SurveyReplication. Arch. Gen. Psychiatry 62, 617–627.

Kessler, R.C., Gruber, M., Hettema, J.M., Hwang, I., Sampson, N., Yonkers, K.A., 2008.Comorbid major depression and generalized anxiety disorders in the NationalComorbidity Survey follow-up. Psychol. Med. 38, 365–374.

Keyl, P.M., Eaton, W.W., 1990. Risk factors for the onset of panic disorder and otherpanic attacks in a prospective, population-based study. Am. J. Epidemiol. 13,301–311.

King, M., Bottomley, C., Bellón-Saameño, J.A., Torres-Gonzalez, F., !vab, I., Rifel, J.,Maaroos, H.I., Aluoja, A., Geerlings, M.I., Xavier, M., Carraça, I., Vicente, B.,Saldivia, S., Nazareth, I., 2011. An international risk prediction algorithm for theonset of generalized anxiety and panic syndromes in general practice atten-dees: predictA. Psychol. Med. 41, 1625–1639.

Kinley, D.J., Walker, J.R., Enns, M.W., Sareen, J.S., 2011. Panic attacks as a risk forlater psychopathology: results from a nationally representative survey.Depress. Anxiety 28, 412–419.

Klauke, B., Deckert, J., Reif, A., Pauli, P., Domschke, K., 2010. Life events in panicdisorder-an update on “candidate stressors”. Depress. Anxiety 27, 716–730.

Knappe, S., Beesdo-Baum, K., Nocon, A., Wittchen, H.U., 2012. Re-examining thedifferential familial liability of agoraphopia and panic disorder. Depress.Anxiety 29, 931–938.

Lahat, A., Hong, M., Fox, N.A., 2011. Behavioural inhibition: is it a risk factor foranxiety? Int. Rev. Psychiatry 23, 248–257.

Lieb, R., Isensee, B., Höfler, M., Pfister, H., Wittchen, H.U., 2002. Parental majordepression and the risk of depression and other mental disorders in offspring.Arch. Gen. Psychiatry 59, 365–374.

Mergl, R., Seidschek, I., Allgaier, A.K., Möller, H.J., Hegerl, U., Kenke, V., 2007.Depressive, anxiety and somatoform disorders in primary care: prevalence andrecognition. Depress. Anxiety 24, 185–195.

Mirza, I., Jenkins, R., 2004. Risk factors, prevalence, and treatment of anxiety anddepressive disorders in Pakistan: systematic review. BMJ 328, 794–798.

Moffitt, T.E., Caspi, A., Harrington, H., Milne, B.J., Melchior, M., Goldberg, D., Poulton,R., 2007. Generalized anxiety disorder and depression: childhood risk factors ina birth cohort followed to age 32. Psychol. Med. 37, 441–452.

Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., 2009. PRISMA Group, 2009.Preferred reporting items for systematic reviews and meta-analyses: ThePRISMA statement. BMJ 339, b2535.

National Institute for Health and Clinical Excellence (NICE), 2011. GeneralisedAnxiety Disorder and Panic Disorder (With or Without Agoraphobia) in Adults.Management in Primary, Secondary and Community Care. NICE, London.

Pietrzak, R.H., Kinley, J., Afifi, T.O., Enns, M.W., Fawcett, J., Sareen, J., 2013.Subsyndromal depression in the United States: prevalence, course, and riskfor incident psychiatric outcomes. Psychol. Med. 43, 1401–1414.

Rapee, R.M., 2012. Family factors in the development and management of anxietydisorders. Clin. Child Fam. Psychol. Rev. 15, 69–80.

Rothman, K.J., Lash, T.L., Greenland, S., 2013. Modern Epidemiology. Mid-CycleRevision, 3rd ed. Lippincott Williams and Wilkins, Philadelphia.

Rudaz, M., Craske, M.G., Becker, E.S., Ledermann, T., Margraf, J., 2010. Health anxietyand fear of fear in panic disorder and agoraphobia vs. social phobia: aprospective longitudinal study. Depress. Anxiety 27, 404–411.

Schmidt, N.B., Zvolensky, M.J., 2007. Risk factor research and prevention for anxietydisorders: introduction to the special series on risk and prevention of anxietypathology. Behav. Modif. 31, 3–7.

Schoevers, R.A., Deeg, D.J.H., van Tilburg, W., Beekman, A.T.F., 2005. Depression andgeneralized anxiety disorder: Co-occurrence and longitudinal patterns inelderly patterns in elderly patients. Am. J. Geriatr. Psychiatry 13, 1.

Silove, D., Manicavasagar, V., Curtis, J., Blaszczynski, A., 1996. Is early separationanxiety a risk factor for adult panic disorder? A critical review. Compr.Psychiatry 37, 167–179.

Smoller, J.W., Pollack, M.H., Wassertheil-Smoller, S., Jackson, R.D., Oberman, A.,Wong, N.D., Sheps, D., 2007. Panic attacks and risk of incident cardiovascularevents among postmenopausal women in the Women’s Health InitiativeObservational Study. Arch. Gen. Psychiatry 64, 1153–1160.

Somers, J.M., Goldner, E.M., Waraich, P., Hsu, L., 2006. Prevalence and incidencestudies of anxiety disorders: a systematic review of the literature. Can. J.Psychiatry 51, 100–113.

Starfield, B., Shi, L., Macinko, J., 2005. Contribution of primary care to healthsystems and health. Milbank Q. 83, 457–502.

Tanios, C.Y., Abou-Saleh, M.T., Karam, A.N., Salamoun, M.M., Mneimneh, Z.N.,Karam, E.G., 2009. The epidemiology of anxiety disorders in the Arab world:a review. J. Anxiety Disord. 23, 409–419.

Vink, D., Aartsen, M.J., Schoevers, R.A., 2008. Risk factors for anxiety and depressionin the elderly: a review. J. Affect. Disord. 106, 29–44.

Wells, G.A., Shea, B., O’Connell, D., Peterson, J., Welch, V., Losos, M., Tugwell, P., 2011.The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomizedstudies in meta-analyses. !www.ohri.ca/programs/clinical_epidemiology/oxford.asp".

Wong, S.L., Wilczynski, N.L., Haynes, R.B., Ramkissoonsingh, R., Team, Hedges, 2003.Developing optimal search strategies for detecting sound clinical predictionstudies in MEDLINE. AMIA Annu. Symp. Proc., 728–732.

World Health Organization (WHO), 2008. The World Health Report: Primary HealthCare (now more than ever). WHO, Geneva.

Zvolensky, M.J., Schmidt, N.B., Bernstein, A., Keough, M.E., 2006. Risk factor researchand prevention programs for anxiety disorders: a translational research frame-work. Behav. Res. Ther. 44, 1219–1239.

P. Moreno-Peral et al. / Journal of Affective Disorders 168 (2014) 337–348348