Late-Onset Agoraphobia: General Population Incidence and Evidence for a Clinical Subtype

Community-dwelling persons ≥65 years old were recruited by random selection from the 15 electoral rolls of the Montpellier, France, district between March 1999 and February 2001 as part of the ESPRIT study of late-life psychiatric disorders (12). Of the persons contacted, 72.7% agreed to participate in the study. Those who declined to participate were replaced by another person drawn at random from the same electoral division so that each division was equally represented. Those who declined were slightly older and more likely to live alone than those who agreed to participate. Participants were given a baseline examination and re-examined with a detailed psychiatric interview on two further occasions at intervals of 2 and 4 years. Seventy persons diagnosed with dementia at baseline were excluded from the present study. Dementia was diagnosed by a neurologist as part of a standardized examination (details are presented below) and validated by a panel of independent neurologists. Of the 2,189 participants without dementia included in the ESPRIT study, 221 were excluded from the present study because of missing data on agoraphobia or on one of the main sociodemographic variables at baseline. Thus, our cross-sectional analyses were conducted for 1,968 persons. These participants did not differ from those who were excluded because of missing baseline sociodemographic or health variables. The longitudinal analysis was carried out for participants with data available from either follow-up examination (2-year or 4-year). Of the baseline sample, 98 persons (5%) were missing data for both follow-up examinations (38 died, 26 were lost to follow-up, and 34 withdrew). Compared with the participants for whom follow-up data were available, those lost to follow-up were older (p<0.001), had a lower educational level (p=0.02), and experienced more frequent baseline respiratory problems (p<0.001), more cardiovascular and cerebrovascular pathologies (p=0.02), and greater disability (p<0.001). There were no significant differences for baseline depressive symptoms, anxiety, or agoraphobia. Participants with incident cases of dementia (N=48) were retained in the study because they were still community-dwelling and able to undertake the examinations. In cases of doubt, family proxies and general practitioners were consulted. The study protocol was approved by the ethics committee at the University-Hospital of Kremlin-Bicêtre, and participants provided written informed consent.

Participants were required to undergo a half-day examination by a neurologist and a center interviewer (a nurse or psychologist) at the Gui de Chauliac Neurology Hospital (Montpellier, France). Persons unable to come to the study center (7% of participants) were examined in their homes. The procedures carried out are described below.

A sensitivity analysis was conducted to compare participants included in the analyses with those excluded using the chi-square test (2). The chi-square and Fisher’s exact tests were used to compare crude rates for psychiatric comorbidities at baseline and follow-up. The clinical characteristics of the sample of 1,968 participants are described according to the presence of agoraphobia or no presence of agoraphobia at baseline, with p values from a logistic regression analysis adjusted for age, sex, and education. The incidence rate over the follow-up was calculated for 1,175 participants with no history of agoraphobia at baseline and with data available at both follow-up assessments. For new cases, the exact date of onset during the follow-up period was not known, and onset was therefore considered to have occurred midway between the two examinations. Population incidence was estimated by dividing the number of new cases that occurred during the follow-up by the total number of agoraphobia-free years lived by the cohort from baseline, expressed as number of new cases per 1,000 person-years. Longitudinal associations between baseline sociodemographic, health, and lifestyle variables and incident agoraphobia were examined using a mixed logistic model (23) (SAS, version 9.2, NLMIXED procedure; SAS Institute, Cary, N.C.), taking into account the correlation between within-subject responses (repeated evaluations). This model allows the inclusion of participants with incomplete responses across time. The analysis of new-onset cases (which excluded recurrent cases) was thus conducted for a maximum of 1,461 participants (2-year follow-up data available, N=1,376; 4-year follow-up data available, N=1,260) with neither past nor current agoraphobia at baseline and with at least one follow-up examination with complete data on agoraphobia and its principal covariates. Variables with a p value <0.15 in the univariate model were entered into a final multivariate model. A Cox proportional hazards model was used to examine the association between baseline agoraphobia and incident cardiovascular events.

Our prospective study of agoraphobia in a large general population cohort, using a standardized clinical interview based on DSM-IV criteria with clinical validation of cases, suggests that there is a much higher prevalence of the disorder in the elderly than previously suggested by the 1988 Epidemiological Catchment Area study (7) (10.4% compared with 1.4%). A subsequent study conducted in the United Kingdom (24), using DSM-III criteria in a small sample (N=168) of elderly persons from a community care service, found a similarly high rate of agoraphobia (14.9%); however, the rate was lowered to 6.5% when hierarchical exclusion rules were applied. The exclusion rules included comorbid major depression, which if applied to our cohort would have more than halved the baseline prevalence estimate.

Over the 4-year follow-up period, we detected and validated 132 new cases of agoraphobia (11.2% of the baseline population) without application of hierarchical decision rules, yielding an estimated general population incidence rate of 32 per 1,000 person-years. During follow-up, these new cases were also found to be associated with incidence of other anxiety disorders (principally generalized anxiety disorder) and suicidal ideation. Although agoraphobia was seen to be commonly preceded by depression, new cases of agoraphobia were not accompanied by incident depression, suggesting that depression is a risk factor and not a consequence of agoraphobia.

The overall rates we observed for panic disorder (<1%) are similar to those found in the National Comorbidity Survey Replication study (25). The importance of panic attacks as a risk factor for agoraphobia in younger populations (26, 27) has led to the assumption that they are part of the same syndrome, largely based on the behavioral hypothesis that agoraphobia is a conditioned avoidance response to the aversive stimulus of spontaneous panic attacks. According to this model, panic attacks are the primary diagnosis, although other investigators, including the behavioral scientist Marks (28), have questioned the temporal precedence and causal role of panic attacks in the development of agoraphobia. Clinical studies have suggested that late-onset cases of agoraphobia in particular may be characterized by high rates of past major depression rather than panic attacks (8), although this has never been investigated in a prospective epidemiological study. Our findings support this observation, with only two of the 132 persons in the incident case group having a previous or recent episode of panic attacks but the group having more than double the rate of depression found in persons with no agoraphobia at baseline. Overall, our results largely support Flint’s (8) observation that depression is the primary comorbidity and that antidepressants or behavioral therapies, rather than benzodiazepines, should be the treatment of choice in elderly patients with agoraphobia. At baseline, we found that only 6% of persons with agoraphobia had been treated with an antidepressant and 13% with an anxiolytic. These rates were not significantly higher than those for the nonagoraphobia group.

The principal documented risk factors for all phobias are younger age, female sex, living alone, physical illness, other anxiety disorders and depression, not raised by both parents, and cognitive disorder (29–32). All of these risk factors were examined in our study. The odds of incident agoraphobia during follow-up decreased with age at baseline, whereas poorer visuospatial memory, high trait anxiety, and severe depression at baseline were associated with increased odds. On the other hand, we found no significant association with living alone, physical illness, or childhood environment. Interestingly, although a higher prevalence rate of agoraphobia was observed among women at baseline, the slightly greater number of women among new cases occurring at age 65 years or older fell short of statistical significance, and this marginal significance disappeared in the multivariate risk model, probably as a result of confounding by depression and anxiety.

Boldrini et al. (33) found agoraphobia to be specifically associated with spatial learning impairment, and our findings further support their observation, with agoraphobia onset being accompanied by poor performance on a visuospatial memory task but not a verbal fluency test. Poor visuospatial performance has been associated with falls and sensory impairment, which may have confounded the association; however, we observed no association with either sensory impairment or falls.

It has previously been suggested (34) that a past history of any anxiety disorder doubles the probability of both suicidal ideation and suicide attempts. We observed higher rates of suicidal ideation in persons with agoraphobia at baseline (16.6% of cases compared with 9.1% of noncases). This observation was further confirmed prospectively, with new cases of agoraphobia paralleled by an increase in the incidence of suicidal ideation. On the other hand, no association was found between onset of agoraphobia and baseline cardiac pathology. Nor were new cases accompanied by an increase in incidence of cardiac disorder in contrast to previous observations relating to phobic anxiety in general (4, 5). Persons with reduced mobility were also found to be at greater risk, although the significance of this relationship disappeared in a multivariate analysis taking into account competing risk factors.

A weakness of this study was its reliance on retrospective recall by participants of life events and previous episodes of agoraphobia. While the refusal rate was low for an epidemiological study, we cannot exclude bias due to loss at both baseline and follow-up of a more disabled group, which, along with not including institutionalized persons at baseline, may have led to an underestimation of the actual number of cases. Depression and poor physical health are strongly associated in the elderly, and while we eliminated confounding due to all the main causes of chronic and acute illness in our study cohort, we did not examine self-perceived health, which may have had an independent effect. Furthermore, while we have demonstrated the significance of depression as a risk factor for agoraphobia, the study design does not permit us to clarify the causal association between the two. Further research is required to clarify whether there may be a direct relationship based on underlying neurophysiological changes driven by chronic depressive symptoms or whether the association is further moderated by complex interactions implicating genetic and biochemical vulnerability, social learning, and cognitive processes.

On the other hand, this study had a number of strengths. Incident case identification procedures were rigorous, and our well-phenotyped study enabled us to examine a large number of potential confounding factors within a prospective design. Overall, our findings suggest that agoraphobia may be relatively common in the elderly, occurring in more than 10% of community-dwelling persons. Late-onset cases show some clear distinctions from clinical reports of early-onset cases. First, there seems to be no clear sex difference in late-onset cases. Additionally, while early-onset cases have been more frequently associated with panic disorder, this appears not to be the case in late-onset cases, in which severe depression is predominant as a risk factor rather than a consequence of the disorder. The principal risk factors for late-onset agoraphobia are younger age, severe depression, trait anxiety, and poor performance on a test of visuospatial memory. Unlike in other phobias, there does not appear to be increased risk for cardiac pathology, although our 4-year follow-up period may have been too short for any definitive conclusions to be drawn on this issue.

Agoraphobia in this cohort is clearly undertreated, with the treatment preference still being for anxiolytics rather than antidepressants, despite the presence of severe depressive symptoms. Screening for agoraphobia in the elderly is clearly worthwhile, since it is a highly treatable condition that may be masked by depression and overlooked when there is no history of panic attacks. Its association in the elderly with suicidal ideation is of clinical concern because persons with agoraphobia may be less likely to seek assistance. Our finding that agoraphobia in the elderly has a clinically atypical presentation compared with agoraphobia in younger populations is also of relevance to the ongoing process of revision of diagnostic criteria for psychiatric disorders, suggesting the need to consider the possibility of a syndrome subtype in older populations.