A comparison of DSM-5 and DSM-IV agoraphobia in the World Mental Health Surveys.

BACKGROUND: The Diagnostic and Statistical Manual of Mental Disorders, version 5 (DSM-5) definition of agoraphobia (AG) as an independent diagnostic entity makes it timely to re-examine the epidemiology of AG. Study objective was to present representative data on the characteristics of individuals who meet DSM-IV criteria for AG (AG without a history of panic disorder [PD] and PD with AG) but not DSM-5 criteria, DSM-5 but not DSM-IV criteria, or both sets of criteria.
METHODS: Population-based surveys from the World Mental Health Survey Initiative including adult respondents (n = 136,357) from 27 countries across the world. The Composite International Diagnostic Interview was used to assess AG and other disorders.
RESULTS: Lifetime and 12-month prevalence estimates of DSM-5 AG (1.5% and 1.0%) were comparable to DSM-IV (1.4% and 0.9%). Of respondents meeting criteria in either system, 57.1% met criteria in both, while 24.2% met criteria for DSM-5 only and 18.8% for DSM-IV only. Severe role impairment due to AG was reported by a lower proportion of respondents who met criteria only for DSM-IV AG (30.4%) than those with both DSM-5 and DSM-IV AG (44.0%; χ 2 1  = 4.7; P = 0.031). The proportion of cases with any comorbidity was lower among respondents who met criteria only for DSM-IV AG (78.7%) than those who met both sets (92.9%; χ 2 1 = 14.5; P < 0.001).
CONCLUSIONS: This first large survey shows that, compared to the DSM-IV, the DSM-5 identifies a substantial group of new cases with AG, while the prevalence rate remains stable at 1.5%. Severity and comorbidity are higher in individuals meeting DSM-5 AG criteria compared with individuals meeting DSM-IV AG criteria only.

INTRODUCTION

Anxiety disorders are highly prevalent and are major contributors to the burden of disease worldwide (Craske et al., 2017; Murray et al., 2012; Whiteford et al., 2013). Agoraphobia (AG) is one of the least studied anxiety disorders (Asmundson & Asmundson, 2018), and, especially with the introduction of the Diagnostic and Statistical Manual of Mental Disorders, version 5 (DSM‐5), there is a lack of research focusing on AG including the cross‐national epidemiology of AG.

In the Diagnostic and Statistical Manual of Mental Disorders, version 4 (DSM‐IV), unlike the International Classification of Diseases (ICD‐10), AG was not defined as an independent disorder with specific diagnostic criteria, but instead was described as a residual group only to be coded in the presence or absence of panic disorder (PD; i.e., PD with AG [300.21], or AG without a history of PD [300.22]; American Psychiatric Association, 2000). AG was defined as anxiety about being in places or situations from which escape might be difficult (or embarrassing) or in which help may not be available in the event of having a panic attack or panic‐like symptoms. These situations had to be avoided, endured with marked distress or with anxiety about having a panic attack or panic‐like symptoms, or require the presence of a companion (American Psychiatric Association, 2000).

In DSM‐5, which was introduced in 2013, AG is defined as a phobic disorder cued by exposure to agoraphobic situations. As a result, PD (300.01) and AG (300.22) are unlinked (American Psychiatric Association, 2013; Asmundson, Taylor, & Smits, 2014), consistent with the ICD‐10 criteria for AG (F40.0) and PD (F41.0; World Health Organization, 2016). The decision to classify AG as a disorder that exists separately from PD was based on studies indicating that a substantial proportion of individuals with AG do not have PD (Kessler et al., 2006; Wittchen, Gloster, Beesdo‐Baum, Fava, & Craske, 2010), and/or PD or panic attacks do not precede AG (as implied in DSM‐IV; Wittchen et al., 2010, 2008; Wittchen, Reed, & Kessler, 1998). As a result, individuals with AG, but without panic‐like symptoms, received no formal diagnosis while they showed substantial impairment and disability (Wittchen et al., 2010). Further, DSM‐5 criterion A is broadened to a fear or avoidance of situations because of thoughts that escape might be difficult or help might not be available in the event of developing panic‐like symptoms or other incapacitating or embarrassing symptoms (e.g., fear of falling in the elderly; fear of incontinence ; American Psychiatric Association, 2013). This is in line with the ICD criteria, in which AG is understood to be the consequence of a broader range of fears (Kogan et al., 2016; Stein, 2012; Wittchen et al., 2010). To make a better distinction from specific phobia (SP), endorsement of fears from two or more distinct situational domains is required in DSM‐5. Finally, the DSM‐5 criteria for AG are extended compared to the DSM‐IV criteria to make them more comparable to other (anxiety) disorders, for instance by adding persistence and severity requirements (American Psychiatric Association, 2013; Asmundson et al., 2014).

These changes in DSM‐5 call for a re‐examination of epidemiological data on AG. The available information is difficult to evaluate because most studies have examined AG only in individuals without a history of PD (Goodwin et al., 2005). As a result, the prevalence of AG (Goodwin et al., 2005) and the impairment due to AG has likely been underestimated. Furthermore, there is a lack of information on age of onset (AOO; Wittchen et al., 2010), sociodemographic correlates, and comorbidity patterns of AG (Goodwin et al., 2005). Therefore, the aim of the current study is to present and compare data on characteristics of AG according to DSM‐5 and DSM‐IV criteria (AG without a history of PD and PD with AG) from countries in the World Health Organization (WHO) World Mental Health (WMH) Survey Initiative.

METHODS

Survey samples

Data came from 27 surveys administered in low/lower‐middle income countries, upper‐middle income countries, and high‐income countries. A total of 136,357 respondents participated. Interviews were conducted face‐to‐face in respondent homes. Adults were selected based on multistage clustered area probability sampling designs designed to generate samples that were representative of the household populations in the countries. The details of within‐country sampling methods are described in detail elsewhere (Heeringa et al., 2008; Pennell et al., 2008).

Ethics, consent, and permissions

Informed consent was obtained according to protocols endorsed by local Institutional Review Boards.

Measures

Mental disorders

Mental disorders were assessed with the WHO Composite International Diagnostic Interview (CIDI), a fully structured interview administered by trained lay interviewers, which generates diagnoses according to the criteria of the DSM‐IV (Kessler & Ustün, 2004). To reduce respondent burden, interviews were administered in two parts. All respondents completed Part I of the CIDI, assessing core mental disorders. Part II, which assessed other disorders and correlates, was administered to all respondents with any lifetime Part I diagnosis and a probability subsample of other Part I respondents. Part II data were weighted to adjust for the undersampling of Part I noncases so that weighted prevalence estimates in Part II sample are identical to those in Part I sample.

The disorders include anxiety disorders (PD, AG, generalized anxiety disorder [GAD], social anxiety disorder, SP, posttraumatic stress disorder [PTSD],1 separation anxiety disorder), mood disorders (major depressive episode and/or dysthymia, bipolar disorder [I, II, or subthreshold]), disruptive behavior disorders (intermittent explosive disorder, bulimia nervosa, binge eating disorder, oppositional defiant disorder, conduct disorder, attention deficit disorder), and substance use disorders (alcohol abuse and drug abuse, both with or without dependence). These diagnoses have shown generally good concordance with clinical diagnoses based on blinded Structured Clinical Interview (SCID) reappraisal (Haro et al., 2006). The AOO of AG was assessed using special recall probes that have been shown to yield more plausible distributions of AOO of disorders than conventional recall questions (Knäuper, Cannell, Schwarz, Bruce, & Kessler, 1999).

For purposes of the current analysis, DSM‐5 AG diagnoses were generated retrospectively; a series of questions were used to operationalize DSM‐5 AG criteria (see Table S1 for the DSM‐IV and DSM‐5 criteria and corresponding CIDI algorithms). We defined three diagnostic groups: respondents who only met DSM‐IV criteria (AG without a history of PD and PD with AG) “DSM‐IV only AG,” respondents who only met DSM‐5 criteria “DSM‐5 only AG,” and respondents who met both DSM‐5 and DSM‐IV criteria “DSM‐5 with DSM‐IV AG.”

Impairment

Severe role impairment in home management, ability to work, ability to form and maintain close relationships, and social life was assessed with a modified version of the Sheehan Disability Scale (SDS) in respondents with 12‐month AG (Leon, Olfson, Portera, Farber, & Sheehan, 1997). The response scale for each role domain is from 0 to 10. Severe impairment was defined as a score ≥ 7 in at least one specific role domain. Respondents with 12‐month AG were also asked how many days in the past year they were totally unable to work or carry out their normal activities due to their AG (Ormel et al., 2008). Additionally, all Part II respondents were asked how many days in the past 30 days they were totally unable to work or carry out their normal activities because of any physical or mental health problems. Finally, all Part I respondents were asked whether they seriously thought about committing suicide in the past 12 months.

Treatment

Respondents were asked whether they ever saw each of a long list of professionals. Responses were aggregated into treatment in the specialty mental health sector (e.g., psychiatrist/psychologist), general medical sector (e.g., general practitioner), human services sector (e.g., social worker), and complementary and alternative medicine (CAM) sector (e.g., herbalist).

Sociodemographic correlates

Factors considered include gender, age cohorts (18–34, 35–49, 50–64, 65+), education level (low, low‐average, high‐average, high), marital status (married, never married, previously married), and employment status (employed, student, homemaker, retired, other).

Statistical analysis

The actuarial method was used to generate AOO survival curves, and differences in age of onset between DSM‐IV only AG, DSM‐5 only AG, and DSM‐5 with DSM‐IV AG were tested using discrete‐time logistic regression in the subsample with DSM‐IV or DSM‐5 AG. Logistic regression analysis was used to evaluate the significance of differences in role impairment, suicidality, comorbidity, and treatment between DSM‐IV only AG, DSM‐5 only AG, and DSM‐5 with DSM‐IV AG cases, and between AG cases and noncases (where applicable). Logistic regression was also used to compare sociodemographic correlates of DSM‐IV only AG, DSM‐5 only AG, and DSM‐5 with DSM‐IV AG. All analyses were carried out in SAS (9.4). Because the data were clustered and weighted to account for unequal selection probabilities, standard errors were estimated using the Taylor series linearization method (Wolter, 1985) implemented in SUDAAN (11.0.1; Research Triangle Institute, 2002). Significance tests were evaluated using 0.05‐level two‐sided tests.

RESULTS

Prevalence and course of AG

Lifetime and 12‐month prevalence estimates of DSM‐IV AG were 1.4% and 0.9%, while those of DSM‐5 AG were 1.5% and 1.0% (Table 1). Consistent with DSM‐IV criteria, 100% of cases with DSM‐IV AG experienced fear of panic attacks, compared to 70.2% of cases with DSM‐5 AG. Of all respondents with lifetime PD (prevalence of 1.7%), 19.7% met criteria for DSM‐IV AG and 18.2% for DSM‐5 AG (Table S2).

Table 1

Lifetime and 12‐month prevalence and persistence of DSM‐IV agoraphobia (AG), DSM‐5 AG, DSM‐IV only AG, DSM‐5 only AG, and DSM‐5 AG with DSM‐IV AG, with and without lifetime panic disorder (PD; total N = 136,357)

	Lifetime		12 months		Persistence (12 months/lifetime)
	%	SE	%	SE	%	SE
DSM‐IV AG
With PD	0.3	0.0	0.2	0.0	68.0	2.5
Without PD	1.0	0.0	0.7	0.0	65.9	1.5
With PD/total	24.7	1.2	25.3	1.5	–	–
With fear of PA/total	100.0	0.0	100.0	0.0	–	–
Total	1.4	0.0	0.9	0.0	66.4	1.3
DSM‐5 AG
With PD	0.3	0.0	0.2	0.0	69.7	2.7
Without PD	1.2	0.0	0.8	0.0	68.9	1.5
With PD/total	21.4	1.1	21.6	1.2	–	–
With fear of PA/total	70.2	1.3	72.4	1.5	–	–
Total	1.5	0.0	1.0	0.0	69.1	1.3
DSM‐IV only AG
With PD	0.1	0.0	0.0	0.0	53.3	6.8
Without PD	0.3	0.0	0.1	0.0	51.6	3.2
With PD/total	19.3	2.4	19.8	3.6	–	–
With fear of PA/total	100.0	0.0	100.0	0.0	–	–
Total	0.3	0.0	0.2	0.0	51.9 a	2.8
DSM‐5 only AG
With PD	0.0	0.0	0.0	0.0	57.3	9.3
Without PD	0.4	0.0	0.3	0.0	64.8	2.7
With PD/total	9.3 b , c	1.5	8.3 d , e	1.3	–	–
With fear of PA/total	0.0	0.0	0.0	0.0	–	–
Total	0.4	0.0	0.3	0.0	64.1 f , g	2.6
DSM‐5 with DSM‐IV AG
With PD	0.3	0.0	0.2	0.0	71.5	2.6
Without PD	0.8	0.0	0.5	0.0	71.1	1.7
With PD/total	26.5	1.4	26.6	1.6	–	–
With fear of PA/total	100.0	0.0	100.0	0.0	–	–
Total	1.0	0.0	0.7	0.0	71.2	1.5
No AG
	98.2	0.0	98.8	0.0	–	–

Note. DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, version 4; DSM‐5: Diagnostic and Statistical Manual of Mental Disorders, version 5.

Significantly different from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 47.1; P < 0.001).

Significantly different from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 54.3; P < 0.001).

Significantly different from DSM‐IV only AG (χ 2 1 = 20.9; P < 0.001).

Significantly different from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 52.8; P < 0.001).

Significantly different from DSM‐IV only AG (χ 2 1 = 20.2; P < 0.001)..

Significantly different from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 6.2; P = 0.012).

Significantly different from DSM‐IV only AG (χ 2 1 = 18.6; P < 0.001).

Of all respondents with lifetime AG, 57.1% (SE = 1.3) met criteria for both DSM‐5 and DSM‐IV AG, 18.8% (SE = 1.0) only met criteria for DSM‐IV, and 24.2% (SE = 1.1) only met criteria for DSM‐5. DSM‐5 only lifetime cases had a significantly lower lifetime proportion with PD (9.3%) than DSM‐5 with DSM‐IV cases (26.5%; χ 2 1 = 54.3; P < 0.001) and DSM‐IV only cases (19.3%; χ 2 1= 20.9; P < 0.001). Respondents with DSM‐5 with DSM‐IV AG had higher persistence (i.e., 12‐month prevalence among lifetime cases; 71.2%) than DSM‐IV only (51.9%; χ 2 1 = 47.1; P < 0.001) and DSM‐5 only (64.1%; χ 2 1= 6.2; P = 0.012) cases. DSM‐5 only cases also had a significantly higher persistence than DSM‐IV only cases (χ 2 1= 18.6; P < 0.001).

The median AOO of DSM‐5 only AG was 14 years old (interquartile range [IQR]= 9–25), which was significantly lower than the median AOO of DSM‐IV only AG (median = 23, IQR = 13–41) and of DSM‐5 with DSM‐IV AG (median = 21, IQR = 13–39; χ 2 = 21.8–24.4; P < 0.001; Figure S1). The AOO of DSM‐IV only AG and DSM‐5 with DSM‐IV AG did not differ significantly (χ 2 = 0.3; P = 0.55).

Impairment

Severe role impairment in the past 12 months was reported by 30.4% of respondents with DSM‐IV only AG, 43.3% of respondents with DSM‐5 only AG, and 44.0% of respondents with DSM‐5 with DSM‐IV AG, with a significant difference between DSM‐IV only and DSM‐5 with DSM‐IV AG (χ 2 1= 4.7; P = 0.031). Mean number of days out of role in the past year due to AG was also significantly lower among respondents with DSM‐IV only AG (29.9) compared with DSM‐5 with DSM‐IV AG (55.8; χ 2 1= 8.0; P = 0.005), while that for DSM‐5 only AG (40.2) did not differ from DSM‐IV only AG (χ 2 1 = 2.1; P = 0.147) or DSM‐5 with DSM‐IV AG (χ 2 1 = 1.7; P = 0.190). Although suicidal ideation rates were higher for all AG subgroups compared with respondents without AG (1.7%), there were no significant differences in suicidal ideation rates among respondents with DSM‐IV only (10.0%), DSM‐5 only (15.7%), or DSM‐5 with DSM‐IV (15.8%) AG (Table 2).

Table 2

Sheehan impairment in the worst month in the past year, days out of role in the past year due to agoraphobia (AG), days out of role in the past 30 days for any health reason, and 12‐month suicidality

	Among respondents with 12‐month AG				Among all respondents
	Any severe impairment due to AG		Number of days out of role due to AG (past year)		Number of days out of role for any health reason (past 30 days)		12‐month suicidal ideation
Diagnosis	%	SE	Mean	SE	Mean	SE	%	SE
DSM‐IV only AG	30.4 a	4.3	29.9 b	8.7	3.3 c	0.8	10.0 d	2.6
DSM‐5 only AG	43.3	4.0	40.2	9.3	3.6 e	0.6	15.7 f	2.8
DSM‐5 with DSM‐IV AG	44.0	2.1	55.8	4.9	5.0 g	0.4	15.8 h	1.3
No AG	–	–	–	–	1.1	0.0	1.7	0.0
χ 2 2/3 [P‐value]	4.8 [0.091]		8.2 [0.016]		128.5 [<0.001]		493.1 [<0.001]

Note. DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, version 4; DSM‐5: Diagnostic and Statistical Manual of Mental Disorders, version 5.

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 4.7; P = 0.031).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 8.0; P = 0.005).

Differs significantly from noncases (χ 2 1 = 6.1; P = 0.013) and from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 5.8; P = 0.016).

Differs significantly from noncases (χ 2 1 = 44.0; P < 0.001).

Differs significantly from noncases (χ 2 1 = 19.8; P < 0.001) and from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 4.2, P = 0.040).

Differs significantly from noncases (χ 2 1 = 97.5; P < 0.001).

Differs significantly from noncases (χ 2 1 = 103.8; P < 0.001).

Differs significantly from noncases (χ 2 1 = 378.2; P < 0.001).

Comorbidity

Respondents in all AG subgroups reported higher rates of lifetime and 12‐month mental disorder comorbidity compared to respondents without AG, except for 12‐month substance use disorders in the DSM‐IV only AG subgroup (χ 2 1 = 0.7; P = 0.39; Table 3). Comorbidity rates were significantly lower for respondents with DSM‐IV only (3.5–78.7%) than for respondents with DSM‐5 with DSM‐IV AG (8.1–92.9%), except for 12‐month disruptive behavior disorders and 12‐month and lifetime substance use disorders. Respondents with DSM‐5 only AG did not have a significantly lower rate of comorbidity (12.6–88.7%) than respondents with DSM‐5 with DSM‐IV AG for any disorder category except any lifetime disorder (χ 2 1= 4.4; P = 0.036). Also, they reported a significantly higher rate of comorbidity than respondents with DSM‐IV only AG for 12‐month anxiety disorder (χ 2 1 = 6.9; P = 0.009) and lifetime anxiety disorders (χ 2 1= 7.1; P = 0.008).

Table 3

Lifetime and 12‐month comorbidity of agoraphobia (AG) with other (lifetime and 12‐month) mental disorders

	Anxiety disorders		Mood disorders		Disruptive behavior disorders		Substance use disorders		Any disorder
Diagnosis	%	SE	%	SE	%	SE	%	SE	%	SE
Lifetime diagnoses
DSM‐IV only AG	63.1 a , b	3.3	44.6 a , b	3.3	20.3 a , d	3.1	18.7 a	2.6	78.7 a , b	3.2
DSM‐5 only AG	78.9 a , c	2.7	54.8 a	2.9	26.3 a	2.8	31.5 a	2.8	88.7 a , d	1.9
DSM‐5 with DSM‐IV AG	84.1 a	1.5	59.6 a	1.8	30.1 a	2.0	28.3 a	1.6	92.9 a	1.1
No AG	14.6	0.2	13.0	0.2	6.1	0.2	10.3	0.2	29.5	0.3
χ 2 3 [P‐value]	1,272.6 [<0.001]		1,172.5 [<0.001]		338.1 [<0.001]		189.5 [<0.001]		677.9 [<001]
12‐month diagnoses
DSM‐IV only AG	52.6 a , b	5.0	30.0 a , f	4.1	17.3 a	3.7	3.5	1.9	68.5 a , c	5.0
DSM‐5 only AG	74.5 a , e	3.8	39.0 a	3.6	15.9 a	2.9	12.6 a	2.7	80.3 a	3.7
DSM‐5 with DSM‐IV AG	77.9 a	1.9	47.1 a	2.2	16.8 a	1.8	8.1 a	1.1	87.7 a	1.6
No AG	8.9	0.1	5.8	0.1	2.7	0.1	2.5	0.1	15.2	0.2
χ 2 3 [P‐value]	1,177.4 [<0.001]		1,049.4 [<0.001]		205.3 [<0.001]		81.6 [<0.001]		777.0 [<0.001]

Note. Anxiety disorders do not include panic disorder, but any disorder does include panic disorder.

DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, version 4; DSM‐5: Diagnostic and Statistical Manual of Mental Disorders, version 5.

Differs significantly from noncases (χ 2 1= 17.7–858.3; P < 0.001).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 11.1–25.3; P < 0.001).

Differs significantly from DSM‐IV only AG (χ 2 1 = 7.1; P = 0.008).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 4.4; P = 0.036).

Differs significantly from DSM‐IV only AG (χ 2 1 = 6.9 P = 0.009).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 9.0; P = 0.003).

Treatment

Respondents in all AG subgroups were more likely to receive any lifetime treatment (31.0–52.8%) than respondents without AG (8.4%), but respondents with DSM‐IV only AG were significantly less likely than respondents with DSM‐5 with DSM‐IV AG to receive any lifetime treatment (Table 4). They were also less likely than respondents with DSM‐5 with DSM‐IV AG to receive 12‐month specialty mental health care or any 12‐month treatment but not less likely to receive 12‐month general medical care, human services, or CAM treatment. Respondents with DSM‐5 only AG were significantly less likely to receive any type of treatment than respondents with DSM‐5 with DSM‐IV AG with the exception of human services or CAM treatment. There were no significant differences in treatment rates between respondents with DSM‐IV only AG and those with DSM‐5 only AG, with the exception of lifetime human services (χ 2 1 = 5.5; P = 0.019).

Table 4

Lifetime and 12‐month treatment rates for people with lifetime and 12‐month agoraphobia (AG) or no AG

	Specialty mental health		General medical		Human services		CAM		Any treatment
Diagnosis	%	SE	%	SE	%	SE	%	SE	%	SE
Lifetime diagnoses
DSM‐IV only AG	37.6 a , b	2.8	38.7 a , c	3.1	5.7 d , e	1.2	9.0 a , g	1.7	54.7 a , h	3.2
DSM‐5 only AG	46.1 a , b	2.8	39.4 a , b	2.5	13.6 a , f	2.1	15.5 a	2.0	62.6 a , b	2.6
DSM‐5 with DSM‐IV AG	56.6 a	1.7	54.9 a	1.6	11.7 a	1.0	19.2 a	1.1	72.7 a	1.5
No AG	14.6	0.2	12.3	0.1	2.6	0.1	2.9	0.1	22.6	0.2
χ 2 3 [P‐value]	945.5 [<0.001]		1,094.7 [<0.001]		200.3 [<0.001]		661.2 [<0.001]		1,089.3 [<0.001]
12‐month diagnoses
DSM‐IV only AG	13.5 a , g	2.9	20.9 a	3.4	2.1	1.1	1.7	1.2	31.0 a , i	4.0
DSM‐5 only AG	20.5 a , h	2.9	21.7 a , b	2.4	4.8 a	1.3	5.9 a	1.6	35.0 a , b	3.1
DSM‐5 with DSM‐IV AG	30.7 a	1.7	35.7 a	1.8	5.6 a	0.8	6.9 a	0.8	52.8 a	1.9
No AG	3.9	0.1	4.8	0.1	0.9	0.0	1.0	0.0	8.4	0.1
χ 2 3 [P‐value]	811.2 [< 001]		779.1 [<0.001]		139.5 [<0.001]		200.5 [<0.001]		1,077.5 [<0.001]

Note. DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, version 4; DSM‐5: Diagnostic and Statistical Manual of Mental Disorders, version 5.

Differs significantly from noncases (χ 2 1 = 23.0–865.3; P < 0.001).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 12.1–36.3; P < 0.001).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 7.6; P = 0.006).

Differs significantly from noncases (χ 2 1 = 8.0; P = 0.005).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 5.1; P = 0.024).

Differs significantly from DSM‐IV only AG (χ 2 1 = 5.5; P = 0.019).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 9.0; P = 0.003).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 9.2 – 9.4; P = 0.002).

Differs significantly from DSM‐5 AG with DSM‐IV AG (χ 2 1 = 8.4; P = 0.004).

Sociodemographic correlates

Younger age, female gender, lower education, not being married, and unemployment were associated with 12‐month and lifetime AG, although some associations did not reach statistical significance for one or more of the diagnostic groups (Table 5). As a group, the ORs for age cohort for DSM‐5 only AG differed from the ORs for DSM‐5 with DSM‐IV AG, yet P‐values for comparing ORs between specific age groups were not statistically significant. Compared to high education, odds for respondents with low, low‐average, and high‐average education, were particularly high for DSM‐IV only AG.

Table 5

Multivariate analysis of sociodemographic correlates of 12‐month and lifetime old DSM‐IV only agoraphobia (AG), DSM‐5 only AG, or DSM‐5 with DSM‐IV AG

		12‐month diagnosis						Lifetime diagnosis
		DSM‐IV only AG		DSM‐5 only AG		DSM‐5 with DSM‐IV AG		DSM‐IV only AG		DSM‐5 only AG		DSM‐5 with DSM‐IV AG
Correlates	Levels	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Gender	Female	2.2	(1.4–3.5)	2.0	(1.4–2.7)	1.7	(1.4–2.1)	2.5	(1.9–3.3)	1.9	(1.5–2.4)	1.8	(1.6–2.1)
	Male	1.0		1.0		1.0		1.0		1.0		1.0
	X 2 1 [P‐value]	12.5	[<0.001]	17.0	[<0.001]	31.5	[<0.001]	38.2	[<0.001]	25.4	[<0.001]	61.2	[<0.001]
Age cohort	18–34	1.6	(0.7–3.8)	2.6 a	(1.3–5.1)	3.6	(2.2–5.9)	1.8	(0.9–3.6)	2.9 d	(1.7–5.2)	3.2	(2.2–4.8)
	35–49	2.5	(1.2–5.4)	2.6 a	(1.3–5.1)	5.4	(3.3–8.7)	2.5	(1.4–4.5)	2.9 d	(1.7–5.1)	4.7	(3.2–7.0)
	50–64	2.0	(1.0–4.0)	1.9 a	(1.0–3.3)	3.8	(2.4–6.0)	1.8	(1.0–3.1)	2.0 d	(1.2–3.2)	3.4	(2.3–4.8)
	65+	1.0		1.0		1.0		1.0		1.0		1.0
	X 2 3 [P‐value]	7.9	[0.049]	7.6	[0.055]	53.4	[<0.001]	11.0	[0.012]	15.3	[0.002]	65.6	[<0.001]
Education	Low	2.4	(1.3–4.3)	1.7 b	(1.1–2.7)	2.2	(1.6–3.0)	2.5 c	(1.7–3.7)	1.7	(1.2–2.5)	1.9	(1.4–2.4)
	Low‐average	2.8	(1.6–4.7)	1.4 b	(0.9–2.0)	1.8	(1.4–2.4)	2.3 c	(1.6–3.3)	1.5	(1.1–2.0)	1.6	(1.3–2.0)
	High‐average	2.3	(1.4–3.8)	1.7 b	(1.2–2.6)	1.3	(1.0–1.6)	2.3 c	(1.6–3.4)	1.6	(1.2–2.2)	1.3	(1.1–1.6)
	High	1.0		1.0		1.0		1.0		1.0		1.0
	X 2 3 [P‐value]	15.9	[0.001]	9.1	[0.027]	32.3	[<0.001]	27.5	[<0.001]	12.2	[0.007]	28.9	[<0.001]
Marriage	Never married	1.7	(1.0–3.0)	1.4	(1.0–1.9)	1.4	(1.1–1.7)	1.3	(0.9–2.0)	1.3	(1.0–1.7)	1.4	(1.1–1.7)
	Previously married	1.3	(0.8–2.1)	1.5	(1.1–2.2)	1.6	(1.3–1.9)	1.1	(0.8–1.7)	1.3	(1.0–1.7)	1.6	(1.3–1.9)
	Currently married	1.0		1.0		1.0		1.0		1.0		1.0
	X 2 2 [P‐value]	4.1	[0.132]	7.3	[0.026]	25.9	[<0.001]	2.5	[0.285]	5.8	[0.055]	33.8	[<0.001]
Employment status	Student	1.6	(0.6–4.1)	1.2	(0.6–2.3)	1.3	(0.8–1.9)	1.2	(0.6–2.4)	1.1	(0.6–1.9)	1.2	(0.8–1.7)
	Homemaker	1.5	(0.9–2.4)	1.5	(1.0–2.3)	1.9	(1.5–2.4)	1.2	(0.8–1.6)	1.4	(1.0–1.9)	1.8	(1.4–2.1)
	Retired	2.1	(1.0–4.2)	0.9	(0.5–1.8)	1.4	(0.9–2.2)	1.6	(0.9–2.8)	1.0	(0.6–1.8)	1.4	(1.0–1.9)
	Other	1.6	(0.9–2.7)	2.6	(1.7–3.9)	2.8	(2.3–3.5)	1.4	(0.9–2.1)	1.8	(1.2–2.5)	2.5	(2.1–3.0)
	Employed	1.0		1.0		1.0		1.0		1.0		1.0
	X 24 [P‐value]	6.5	[0.164]	21.0	[<0.001]	104.4	[<0.001]	4.8	[.305]	10.8	[0.029]	114.5	[<0.001]

Note. DSM‐IV: Diagnostic and Statistical Manual of Mental Disorders, version 4; DSM‐5: Diagnostic and Statistical Manual of Mental Disorders, version 5.

ORs are significantly different from ORs for DSM‐5 with DSM‐IV AG (X 2 3 [P‐value] for group =8.6 [0.035], P‐value for comparing ORs for age 18–34 = 0.44, age 35–49 = 0.07, age 50–64 = 0.06).

ORs are significantly different from ORs for DSM‐IV only (X 2 3 [P‐value] for group =8.3 [0.040], P‐value for comparing OR for low education =0.06, low‐average education =0.007, high‐average education =0.35).

ORs are significantly different from ORs for DSM‐5 with DSM‐IV AG (X 2 3 [P‐value] for group =9.6 [0.022], P‐value for comparing OR for low education =0.014, for low‐average education =0.012, and for high‐average education =0.005).

ORs are significantly different from ORs for DSM‐5 with DSM‐IV AG (X 2 3 [P‐value] for group =8.9 [0.030], P‐value for comparing ORs for age 18–34 = 0.85, age 35–49 = 0.18 and for age 50–64 = 0.11).

DISCUSSION

This study is the first to present representative data for DSM‐5 compared to DSM‐IV AG (AG without a history of PD and PD with AG) from countries across the world. Lifetime and 12‐month prevalence estimates of DSM‐5 AG were 1.5% and 1.0%, while those of DSM‐IV AG were 1.4% and 0.9%. Hence, there was no marked shift in AG prevalence from DSM‐IV to DSM‐5. However, only 57.1% of respondents with AG met criteria in both diagnostic systems, with 18.8% of AG cases meeting criteria only for DSM‐IV and 24.2% meeting criteria only for DSM‐5. Compared with DSM‐IV AG, DSM‐5 AG was characterized by a higher persistence of AG and higher rates of severe role impairments, treatment‐seeking, and mental disorder comorbidity.

Several sociodemographic correlates of AG were identified, consistent with other studies of AG and PD (Andrews & Slade, 2002; de Jonge et al., 2016). Although some quantitative differences were found for associations between sociodemographic factors and diagnostic subgroups, the general patterns were comparable, with younger age, female gender, lower education, not being married, and unemployment being associated with AG.

Although our prevalence rates are comparable to the reported prevalence of 12‐month DSM‐IV AG without a history of PD across European countries (1.3%; Goodwin et al., 2005), they are in contrast with the extremely low prevalence rate of 0.05% for 12‐month DSM‐IV AG without a history of PD found in a nationally representative study in the United States, which may (partly) result from the exclusion of lifetime PD cases and a clinical significance criterion used in that study (Grant et al., 2006). Although our 12‐month prevalence rate of DSM‐IV AG increased by including AG with PD, it is interesting to note that 75% of respondents with 12‐month DSM‐IV AG did not have a history of PD. This percentage is comparable to that found for 12‐month AG in an Australian survey, in which 64% did not have a history of PD (Andrews & Slade, 2002), and clearly supports the decision made in DSM‐5 to separate AG from PD.

The relatively stable prevalence rates for DSM‐IV and DSM‐5 AG may result from a broadening of the range of fears beyond a fear of panic attacks or panic‐like symptoms in DSM‐5 on the one hand, whereas increasing the strictness of the (severity) criteria on the other hand. None of the respondents with DSM‐5 only AG reported fear of panic attacks in the feared situations. In addition, the percentage of respondents reporting severe role impairment, and the mean number of days out of role in the past year due to AG, were significantly lower for respondents with DSM‐IV only compared with respondents with 12‐month DSM‐5 with DSM‐IV AG. However, the mean number of days out of role in the past 30 days for any health reason, and 12‐month suicidal ideation, were increased for all AG subgroups compared with respondents without AG. In addition, respondents with DSM‐IV only AG also had increased treatment and mental comorbidity rates compared with respondents without AG. Although these results indicate that individuals who meet DSM‐IV but not DSM‐5 AG criteria also suffer from clinically significant symptoms including impairment, this suffering may for a large part be the result of comorbid mental disorders. However, we cannot exclude the possibility that a small group of individuals who experience severe distress or impairment are left without a diagnosis as a result of the changes in DSM‐5, but the population prevalence rate of lifetime DSM‐IV only AG without mental disorder comorbidity is very low, namely 0.08%.

We found a particularly low median AOO for DSM‐5 only AG. This result adds to previous studies that showed that AG is not a mere consequence of PD (Wittchen et al., 2008, 1998) and that the median AOO of AG appears to be lower than the AOO of PD (Wittchen et al., 1998). The current study therefore confirms that DSM‐5 AG is more comparable to other phobic disorders, which have early onsets as well (Kessler et al., 2005; Stein et al., 2017). Identifying individuals who suffer from AG at a younger age may lead to earlier AG focused treatment and thereby has the potential to prevent disorder progression and development of comorbidity (Jones, 2013; Kessler et al., 2005).

One of the reasons for the controversy whether AG can be regarded as a disorder that is independent of PD (Grant et al., 2006; Wittchen et al., 2008, 1998) is that AG without a history of PD is not typically seen in clinical studies and it has been suggested that many individuals diagnosed with AG in epidemiological studies actually suffer from SP (Horwath, Lish, Johnson, Hornig, & Weissman, 1993). However, another study validated the existence of AG in the absence of panic attacks in a survey through a re‐evaluation with structured interviews (Faravelli, Cosci, Rotella, Faravelli, & Dell’Osso, 2008). A better distinction between SP and AG probably reflects better measurements of AG in recent surveys (Andrews & Slade, 2002; Wittchen et al., 2010). For example, to meet criteria for AG in this study, agoraphobic anxiety had to apply to at least two situations. The low percentage of AG cases seen in clinical care might also be the result of these individuals being less likely to seek help (Andrews & Slade, 2002; Wittchen et al., 1998), potentially as a result of avoidance behavior (Andrews & Slade, 2002). Indeed, the percentage of individuals with AG seeking treatment was lower compared with rates for individuals with PD in the WMH surveys (de Jonge, Roest, Lim, Levinson, & Scott, 2018; Roest et al., 2018).

A previous review article concluded that AG should be seen as a disorder independent from PD based on data from community samples on prevalence rates, temporal relationships of AG, panic attacks, and PD, and impairment associated with AG without panic attacks (Wittchen et al., 2010). Greene and Eaton (2016) also argued that AG is a diagnostic entity that should be separated from PD. By examining multivariate comorbidity patterns of PD and AG, the authors showed that AG could be categorized as a fear disorder, whereas PD is more strongly related to distress disorders, such as depression (Greene & Eaton, 2016). Yet, evidence is lacking for specific genetic underpinnings of AG (Wittchen et al., 2010) and for differential treatment effects; the latter may result from a lack of treatment studies focusing on patients with AG without PD (Bandelow, 2017; Wittchen et al., 2010). To our knowledge only one other study examined the effect of changes from DSM‐IV to DSM‐5 criteria for AG. This study examined the effect of the more stringent criterion A in DSM‐5 (endorsement of fears from multiple distinct situational domains) on the prevalence rate of AG in children and adolescents seeking anxiety treatment (Cornacchio, Chou, Sacks, Pincus, & Comer, 2015). Authors concluded that this adaptation may be too strict for youth, because a substantial proportion (25%) of individuals no longer met criteria for AG, despite being more similar regarding symptomatology and impairment, to individuals who met the new AG criteria than to individuals with SP (Cornacchio et al., 2015). Whether and how the changes in DSM‐5 criteria will affect clinical treatment is yet unclear (Bandelow, 2017); however, additional research into DSM‐5 AG in general (Wittchen et al., 2010), and specifically the treatment of DSM‐5 AG without PD or panic attacks is warranted, especially given the high persistence of DSM‐5 AG shown in the current study.

A strength of the current study is the use of data from the WMH surveys. The WHO WMH Survey Initiative provides a unique opportunity to examine the cross‐national epidemiology of AG because it included data from countries with different income ranges. The numbers of respondents were large enough to examine both lifetime and 12‐month AG and compare diagnostic groups based on DSM‐IV and DSM‐5 criteria, although not large enough to additionally examine differences between countries. Another strength is that the surveys used a common protocol and instrument to assess AG separately from PD. However, this study also has a number of limitations. First, although we examined AG using both DSM‐IV and DSM‐5 criteria, the algorithm was modified in retrospect, as was the case in other reports on DSM‐5 criteria for GAD, PTSD, and substance use disorder in the WMH surveys (Ruscio et al., 2017; Slade et al., 2016; Stein et al., 2014). As a result, the CIDI questions do not match all DSM‐5 criteria perfectly. Although this could have caused misclassification in some cases concerning the DSM‐5 diagnosis of AG, we do not expect this to have a large impact on the comparisons between results for the AG subgroups, since the criteria based on the CIDI questions were sometimes less and sometimes more strict than original DSM‐5 criteria. Second, because the data are cross‐sectional, AOO is reported retrospectively and the indicator of AG persistence is a proxy indicator.

In conclusion, this study is the first to investigate and compare the epidemiology of AG according to the DSM‐5 and the DSM‐IV. Results show that the DSM‐5 criteria may be an improvement over the DSM‐IV criteria as the DSM‐5 identifies individuals with a higher disorder persistence, severity, and comorbidity and help‐seeking rates, whereas the global lifetime and 12‐month prevalence rates remained relatively constant.

CONFLICT OF INTERESTS

In the past 3 years, Dr. Stein has received research grants and/or consultancy honoraria from AMBRF/Foundation for Alcohol Research, Biocodex, Cipla, Lundbeck, National Responsible Gambling Foundation, Novartis, Servier, and Sun. Dr. Demyttenaere has served on advisory boards for Eli Lilly, Lundbeck, Johnson&Johnson, Servier, Boehringer Ingelheim, Livanova and has research grants from Eli Lilly, foundation “ga voor geluk,” Fonds voor Wetenschappelijk Onderzoek Vlaanderen.

In the past 3 years, Dr. Kessler received support for his epidemiological studies from Sanofi Aventis; was a consultant for Johnson & Johnson Wellness and Prevention, Sage Pharmaceuticals, Shire, Takeda; and served on an advisory board for the Johnson & Johnson Services Inc. Lake Nona Life Project. Kessler is a co‐owner of DataStat, Inc., a market research firm that carries out healthcare research. Dr. Haro reports personal fees from Roche, Lundbeck, Eli Lilly and Otsuka, outside the submitted work.

Authors Roest, de Vries, de Jonge, Wittchen, Lim, Adamowski, Carmen Viana, Florescu, Kawakami, Slade, Torres, Posada‐Villa, Lépine, Al‐Hamzawi, Levinson, de Girolamo, Karam, Elena Medina Mora, Gureje, O’Neill, Hu, Piazza, Miguel Caldas‐de‐Almeida, Navarro‐Mateu, Bromet, and Scott do not have conflict of interests to report.

Supplementary information

Click here for additional data file.

Supplementary information

Click here for additional data file.

Supplementary information

Click here for additional data file.

Supplementary information

Click here for additional data file.