Separation Anxiety and Measures of Suicide Risk Among Patients With Mood and Anxiety Disorders


Background: Separation anxiety disorder may be important when considering risk of suicide. The aim of this study was to examine the association between both childhood and adult separation anxiety (disorder) and measures of suicide risk in a large cohort of outpatients with anxiety and mood disorders.

Methods: The sample included 509 consecutive adult psychiatric outpatients with DSM-IV mood disorders or anxiety disorders as a principal diagnosis recruited at the Department of Psychiatry, University of Pisa, Italy, between 2015 and 2018. Suicide risk was evaluated by the Hamilton Depression Rating Scale (HDRS) item 3. Patients were classified in 2 groups: those with a score ≥ 1 and those with a score of 0 on HDRS item 3. Suicide risk was also evaluated by specific items within the Mood Spectrum, Self-Report (MOODS-SR), a questionnaire evaluating lifetime suicidal symptoms. Separation anxiety (disorder) was assessed based on the Structured Interview for Separation Anxiety Symptoms in Adulthood/Childhood (SCI-SAS-A/C), the Separation Anxiety Symptom Inventory (SASI), and the Adult Separation Anxiety Scale (ASA-27).

Results: Of the 509 patients, 97 had an HDRS item 3 score ≥ 1, and 412 had a score of 0. Adult separation anxiety disorder was more frequent among individuals who had suicidal thoughts (53.6%) than those who did not (39.6%) (P = .01). Dimensional separation anxiety symptoms on all scales were elevated in patients with suicidality when compared to patients without (SASI: P = .02; SCI-SAS-C: P < .001; SCI-SAS-A: P < .001; ASA-27: P = .002). Logistic regression found that adult separation anxiety disorder (odds ratio [OR] = 1.86, 95% CI = 1.16–2.97), major depression (OR = 7.13, 95% CI = 3.18–15.97), bipolar I disorder (8.15, 95% CI = 3.34–19.90), and bipolar II disorder (OR = 8.16, 95% CI = 3.50–19.05) predicted suicidal thoughts. Linear regression found that depression (P = .001) and ASA-27 separation anxiety (P = .001) significantly predicted lifetime suicide risk. Mediation analysis found that separation anxiety significantly mediated the association between depression and suicide risk.

Conclusions: This study indicates a substantial role of separation anxiety in predicting suicidal thoughts, both as state-related symptoms (evaluated by HDRS item 3) and as longitudinal dimensional symptoms (as evaluated by MOODS-SR). Greater understanding of the influence of separation anxiety in patients with affective disorders may encourage personalized interventions for reducing suicide risk.

J Clin Psychiatry 2021;82(2):20m13299

To cite: Pini S, Abelli M, Costa B, et al. Separation anxiety and measures of suicide risk among patients with mood and anxiety disorders. J Clin Psychiatry. 2021;82(2):20m13299.
To share:

© Copyright 2021 Physicians Postgraduate Press, Inc.

aDepartment of Clinical and Experimental Medicine, Section of Psychiatry, University of Pisa, Pisa, Italy
bDepartment of Pharmacy, University of Pisa, Pisa, Italy
cDepartment of Psychiatry and Psychotherapy, Medical Center—University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany
dClinical and Experimental Sciences, University of Southampton Faculty of Medicine, Southampton, United Kingdom
eUniversity Department of Psychiatry and Mental Health, University of Cape Town, Cape Town, South Africa
fDepartment of Psychiatry and Psychotherapy, University Medical Center, Gottingen, Germany
gCenter for Basics in NeuroModulation, Faculty of Medicine, University of Freiburg, Freiburg, Germany
*Corresponding author: Stefano Pini, MD, Santa Chiara Hospital, Department of Psychiatry, Building 4, 67 Via Roma, 56126, Pisa, Italy (


There is a close link between suicidal behaviors and interpersonal difficulties, extending beyond the traditional approach of comprehending suicide as a phenomenon mainly related to depression.1 Pompili2 has argued that confining suicidal ideation to the realm of depressive symptoms is detrimental when framing suicidality in a particular individual. Within this framework, “separation anxiety” might indicate a cluster of factors related to suicide. Feelings of abandonment and severe separation anxiety symptoms are not rare in patients with mood and anxiety disorders in connection with situations, either objective or perceived, that threaten their bond and proximity to significant others.3–5 Using a large epidemiologic database (National Comorbidity Survey Replication, N = 5,692, subjects aged > 18 years), Palitsky et al6 found that subjects with an insecure adult attachment style had an increased likelihood of suicidal thoughts and suicide attempts and suggested an association between individual abnormal attachment sensitivity (a psychological construct close to separation anxiety) and suicide. Suicidal ideation or suicide attempts may be associated with disturbances in attachment, which may lead not only to a devastating experience of losing the feeling of interdependence and closeness but also to a rejection of life itself.7

Adult separation anxiety disorder may be a key factor in understanding the relationship between individual attachment sensitivity to separation and suicidality. When separation distress becomes excessive, prolonged, developmentally inappropriate, or impairing, separation anxiety disorder can be diagnosed. Although separation anxiety disorder was traditionally a diagnosis reserved for children and adolescents, DSM-58 removed the 18-year age-at-onset restriction on diagnosis because studies had found a later onset to be common: its onset spans the life course, even without a history of separation anxiety in childhood.9,10 The importance of adult-onset separation anxiety disorder is indicated by a lifetime prevalence of 6.6% in the general population11 and a prevalence of 20%–40% in patients with mood and anxiety disorders.12,13 The relationship between separation anxiety and suicidality has not been explored extensively. One study found an association between separation anxiety disorder and increased risk of suicidal behaviors (odds ratio = 3.28) in a prospective study of 500 Indian adolescents in a rural community.14 Another study reported an association between severity of separation anxiety symptoms and suicidal ideation in a small sample (N = 31) of patients with social anxiety disorder, although the observed association was dependent on comorbidity with major depression and avoidant personality disorder.15

The aim of the present study was to examine the association between separation anxiety (disorder) and suicidal thoughts and attempts in a large cohort of outpatients with anxiety and mood disorders.



Participants were recruited between 2015 and 2018 within the adult psychiatric outpatient clinic at the Department of Psychiatry, University of Pisa, Italy. Patients with psychotic disorders or substance use disorder were excluded. The study sample comprised consecutively recruited adult patients with DSM-IV mood disorders or anxiety disorders (including obsessive-compulsive disorder) as principal diagnosis.16 All participants were assessed with the Structured Clinical Interview for DSM-IV (SCID-I)17 by experienced residents in psychiatry to establish DSM-IV Axis I primary diagnosis and comorbidity. Current anxiety symptoms were evaluated by the Hamilton Anxiety Rating Scale (HARS),18 and current depressive symptomatology was assessed using the 21-item Hamilton Depression Rating Scale (HDRS).19 According to previous research on HDRS, a total score based on the first 17 items was used in the present study.20,21 The Mood Spectrum, Self-Report questionnaire (MOODS-SR, lifetime version), an instrument comprising 161 items, was used to assess a broad spectrum of lifetime mood symptoms, including suicidal ideation and behavior (see below). Its dimensional nature allows for the identification of mild and subthreshold manifestations that may feature subclinical, prodromal, residual, or atypical clinical pictures.22,23 Items are grouped in 7 domains. The questionnaire has good internal consistency (the Kuder-Richardson coefficient ranging from 0.79 to 0.92 across domains). Interviews and assessments performed by experienced residents in psychiatry as well as self‐report questionnaires were completed within 2 days of the initial presentation of the participants.

The study was carried out in accordance with the Declaration of Helsinki and with the approval of the University of Pisa Ethical Committee. All participants were informed of the nature of study procedures and provided written informed consent.

Assessment of Suicidality Using HDRS Item 3

Suicide-related symptoms were rated using HDRS item 3: (0) absent, (1) feels life is not worth living, (2) wishes he/she were dead or any thoughts of possible death to self, (3) suicide ideas or gesture, (4) attempts at suicide (any serious attempt rates 4). Subjects were classified as suicidal if they were rated with a score of ≥ 1 on item 3.

Assessment of Suicidality Using MOODS-SR (Lifetime Version)

Suicidal ideation and behaviors were evaluated by MOODS-SR items 102 through 107, which reflect suicidal ideation (102: “In the course of your life, including when you were a child, have you ever had periods of at least 3 days in which you thought that life was not worth living?”; 103: “[…] you hoped that you would not wake up in the morning, or that you would die in an accident or from something like a heart attack or a stroke?”; 104: “[…] you wanted to die or hurt yourself?”; 105: “If you felt that you wanted to die, did you have a specific plan to hurt or kill yourself?”; 106: “Did you actually try to kill yourself”; 107: “If you tried to kill yourself, did you require medical attention?”). The sum of positive answers to these items was labeled as a “suicidality score.”23–25

Assessment of Childhood and Adulthood Separation Anxiety

Adult and childhood history of separation anxiety disorder was assessed using the Structured Clinical Interview for Separation Anxiety Symptoms (SCI-SAS).26 This instrument evaluates each of the 8 DSM-IV criteria of separation anxiety, separately for childhood (SCI-SAS-C) and adult symptoms (SCI-SAS-A). Each item was scored as 0 (not at all), 1 (sometimes), 2 (often), or ? (do not recall). In accordance with DSM-IV guidelines, fulfillment of 3 or more of the 8 criterion symptoms (symptoms rated as often) was used as a threshold to determine categorical (yes⁄no) diagnosis of separation anxiety disorder in childhood and adulthood. In addition, criteria B (ie, duration of at least 4 weeks) and C (ie, the disturbance causes clinically significant distress or impairment in social, academic, and occupational or other important areas of functioning) were required. Scores on the 8 items of each subscale were summed to produce a continuous measure of separation anxiety symptoms (range for each subscale, 0–16). In the present study, the duration criterion as suggested in the DSM-5, which is typically 6 months or longer in adults as a general guideline, was respected, with allowance for some degree of flexibility (criterion B).

Separation anxiety was further evaluated using 2 self-report questionnaires. The Separation Anxiety Symptom Inventory (SASI)27 is a 15-item self-report inventory developed to assess adult report of separation anxiety experienced during childhood. A total score is obtained by adding single item scores, followed by a square root transformation: as examples, a raw score of 16 generates a transformed score of 4, whereas a score of 9 transforms into a score of 3. This questionnaire has good psychometric properties, including high internal consistency and test-retest reliability. The Adult Separation Anxiety Questionnaire (ASA-27)28 is a 27-item inventory that rates symptoms of adult separation anxiety. Participants responded to items on a 4-point frequency scale, ranging from “This has never happened” to “This happens very often.” Single item scores (0–3) are added to yield a total score ranging from 0 to 81. The measure has shown a high level of internal consistency (Cronbach α = .89) and test-retest reliability (r = 0.86, P < .001).28 A total score of 22 or higher provides a threshold for SAD caseness based on a previous calibration of the measure against a structured clinical interview. This scale has satisfactory concurrent validity with clinical assessments of adult separation anxiety disorder.28

Statistical Analyses

The mean values (± SD) of continuous variables such as age and psychometric scores were compared between groups by means of t tests. Comparisons of categorical variables between groups were conducted using χ2 tests. Bivariate correlations between dimensional measures of adult separation anxiety and HDRS total score and MOODS-SR suicidal items sum score were analyzed by Pearson test. The association between categorical separation anxiety disorder and suicidality was also evaluated by binary logistic regression with odds ratios with 95% confidence intervals. Six predictors (lifetime diagnosis of major depression, bipolar I disorder, bipolar II disorder, panic disorder, adult separation anxiety disorder, and gender) were included in the model, using the backward conditional method. The relationships between suicidality, as assessed by the MOODS-SR, and separation anxiety as well as HDRS total score and age were investigated by using a linear regression model with multicollinearity statistics applied. α was set at P < .05. Since both HDRS total score and ASA-27 total scores were significantly associated with MOODS-SR suicidality score using linear regression analysis, a mediation analysis was performed with HDRS total score total score as predictor, MOODS-SR suicidality score as dependent variable, and ASA-27 total score as mediator. The Hayes PROCESS tool was utilized; bootstrap confidence intervals and Sobel test for indirect effect were computed.

Statistical analyses were conducted using SPSS (SPSS Science, Chicago, Illinois; software version 25.0).


The study sample comprised 509 consecutively recruited adult patients with DSM-IV mood or anxiety disorders (including obsessive-compulsive disorder) as principal diagnosis (n = 173: major depression, mean age at onset: 32.62 ± 11.12 years; n = 175: bipolar disorder, mean age at onset: 28.16 ± 9.78 years; n = 158: panic disorder, mean age at onset: 28.78 ± 11.10 years; n = 50: social anxiety disorder, mean age at onset: 18.60 ± 6.7 years; n = 102: obsessive-compulsive disorder, mean age at onset: not available) (see Table 1 for diagnostic details). Within this patient sample, 215 patients were diagnosed with separation anxiety disorder (mean age at onset: 14.79 ± 12.18 years). The distribution of HDRS item 3 scores in the overall sample was as follows: 412 (80.9%) patients scored 0; 62 (12.2%) patients scored 1; 24 (4.7%) patients scored 2; and 11 (2.1%) patients scored 3. Overall, 97 patients (18.9%) had a score of 1 or greater on HDRS item 3 and so were designated as showing “suicidality.” Demographic and diagnostic characteristics of patients with versus without current suicidality based on HDRS item 3 are given in Table 1.

Gender distribution was similar in the 2 groups (females, 74.2% with suicidality vs 63.8% with no suicidality, P = .052). Diagnoses of adult separation anxiety, panic disorder, and bipolar II disorder, as well as lifetime suicidality according to MOODS-SR items 102–107, were more frequent among patients with suicidality based on HDRS item 3 (see Table 1). Total scores on HDRS (19.08 ± 7.3 vs 9.10 ± 6.7, P ≤ .001) and HARS (18.30 ± 9.7 vs 10.38 ± 7.6, P ≤ .001) were significantly higher in the group with suicidality than in the group without.

Separation Anxiety and Current Suicidality (HDRS Item 3)

All measures of adult as well as childhood separation anxiety were significantly elevated in the group of patients with current suicidality based on HDRS item 3 (Table 2).

A backward binary logistic regression model was found to fit the data adequately (Hosmer and Lemeshow χ2 = 11.829, P = .159). Overall, the model was able to correctly predict 80.9% of all cases. Adult separation anxiety, major depression, bipolar I disorder, and bipolar II disorder predicted current suicidality according to HDRS item 3 (Table 3).

Separation Anxiety/Depression and Lifetime Suicidality (MOODS-SR)

Significant positive bivariate correlations were found for ASA-27 with HDRS total score (Pearson correlation = 0.246, P = .001) and with MOODS-SR suicidal dimension (Pearson correlation = 0.223, P = .001). As shown in Table 4, linear regression analysis identified both the HDRS total score (P = .003) and the ASA-27 total score (P = .012) as predictors of lifetime MOODS-SR suicidality. All values of tolerance were around 1, providing a good value versus collinearity.

A mediation analysis (Figure 1) revealed a significant direct effect of HDRS total score on MOODS-SR suicidality (b = 0.0558, P = .014). The HDRS total score also showed a significant indirect effect on MOODS-SR suicidality total score mediated by the ASA-27 total score (b = 0.0089, 95% CI: 0.0001–0.0238) (κ2 = 0.028, 95% CI: 0.004–0.0757), with an indirect effect of 2.8%. The Sobel test confirmed a mediating effect of the ASA-27 total score (b = 0.18, Z = 3.25, P = .001).


The present study is the first, to our knowledge, to explore the impact of childhood as well as adult separation anxiety (disorder) on suicidality in a large sample of patients with mood and anxiety disorders. We found that the categorical diagnosis of adult (but not childhood) separation anxiety disorder—as well as dimensional measures of separation anxiety symptoms both in adulthood and in childhood—were more frequent in individuals with current suicidality compared to those without.

Cross-sectional community and clinical studies have repeatedly demonstrated that anxiety disorders are associated with suicidal ideation, suicide attempts, and completed suicide.29 Among anxiety disorders, panic disorder has received the greatest attention, but it has been questioned whether panic disorder is associated with suicidal behavior after adjusting for other co-occurring anxiety disorders and other mental disorders.30 For instance, one recent meta-analysis found an increased suicidal risk in patients with panic disorder mostly attributable to alcohol dependence, agoraphobia (in the context of panic disorder), and comorbid depression.31 This accords with another recent study reporting suicidal ideation in patients with panic disorder to be driven by comorbid depression and depression severity rather than anxiety measures.32 Also, in another sample of psychiatric outpatients neither anxiety disorders nor depression as categorical disease entities were associated with elevated suicide risk, but risk was associated with symptom dimensions such as reports of worthlessness, crying, and sadness, higher levels of anxious arousal, and negative affect.33 These data are consistent with the finding that in our study the suicidality group was characterized by more severe psychopathology compared to the group without suicidality as indicated by higher scores on the Hamilton scales for depression and anxiety. The role of panic disorder seems to be more complex. Panic disorder was more frequent in the suicidal group (61%) than in the nonsuicidal group (53%); however, its predictive effect on suicidality, as assessed by logistic regression, appeared weaker compared to separation anxiety, major depression, and bipolar disorder. In sum, these data suggest that separation anxiety—when appropriately diagnosed according to DSM-5 criteria—constitutes a risk factor in predicting suicidality independently of and in addition to panic disorder. This notion is in line with the hypotheses of Shear et al11 and Silove and Rees34 that separation anxiety in adulthood should not be exclusively considered an epiphenomenon of the panic-agoraphobia spectrum.34–38

Unlike HDRS item 3, which covers current suicidality, the MOODS-SR explores whether an individual had a period of at least some days across the lifespan in which suicidal symptoms occurred and assesses their severity. The mood spectrum features as evaluated with the MOODS-SR might manifest in “waves” during the lifespan: presenting sometimes together, sometimes alone, sometimes reaching the severity of a manifest disorder, sometimes interfering with other mental disorders or complicating the course of somatic diseases. Higher scores on the factors assessing suicidality delineate a specific subgroup of patients characterized by more severe psychopathology.36 Our linear regression found a predictive association between adult separation anxiety and severity of lifetime suicidal ideation or behaviors. However, mediation analysis showed that besides a direct effect there is also an indirect effect of depression severity on MOODS-SR suicidality score through the ASA-27 score, indicating that separation anxiety may act as an important mediating factor in the relationship between depression and suicidality.


Some limitations must be acknowledged. The study was not originally designed with a focus on suicide; therefore, there are no specific scales to evaluate risk or other aspects of phenomenology of suicidality.37 However, HDRS item 3 scores were rated by expert psychiatrists trained in the use of the HDRS: in assigning a score, both symptom intensity and symptom frequency were taken into account as recommended.38 In previous studies, suicidality was also evaluated by single items contained in questionnaires assessing overall psychopathological symptoms.39–41 Another limitation is that, at present, there are no scales available to assess the intensity of single current separation anxiety symptoms. For all items of the SCI-SAS and the ASA-27, the presence of the symptom has been scored as “absent,” “occasional,” “much of the time,” and “almost all of the time.” Therefore, the relationship between a “state-dependent” positive answer on the HDRS item 3 and separation anxiety should be interpreted with caution.42 In addition, the HDRS assessment by definition refers to the 2 to 3 weeks preceding the assessment. Therefore, scores higher than 1 on the HDRS item 3 do not necessarily imply an enduring proneness to suicide. Also, the lack of patients scoring 4 on the HDRS item 3 in the present sample limits the generalizability of results to a suicidal population. In a similar vein, the MOODS-SR does not make it possible to differentiate the different types of thoughts linked with progression to the suicidal act. This may be of particular relevance to the interpretation of the present data, since, for instance, anticipatory, relief-oriented, and permissive beliefs have been described in patients with suicidal behaviors and warrant further investigation also in the present context.43 Another limitation is that participants had mood or anxiety disorders as a primary diagnosis with high rates of reciprocal comorbidity and without data on chronology in terms of age at onset, not allowing for analyses stratified for anxiety and affective disorders. Also, no subgroup of patients with agoraphobia (without concurrent panic disorder) could be defined. Notwithstanding, panic disorder and mood diagnoses were included in the regression model to control for their individual independent predictive effect on suicidality. Along these lines, it must be acknowledged that in the present study no information was available on Axis II diagnoses. This could have introduced a bias, given previous reports on a close interaction of depressive/anxiety disorders, personality disorders—particularly cluster B and avoidant personality disorder—and suicidal ideation/attempts.15,33,38 Thus, the presently observed association between separation anxiety and suicidality cannot be generalized to patients with psychiatric diagnoses other than affective disorders or to the general population. Also, the present analyses could not be corrected for the potential influence of early trauma. This is of particular relevance in the light of earlier studies demonstrating adverse childhood experiences such as maltreatment to constitute a risk factor for suicidality.44–46 Given that early trauma—but interestingly not separation events per se43,44—has furthermore been suggested to increase separation anxiety,47–49 there might be a mediating effect of separation anxiety, similar to what was presently found for depression, on the link between childhood trauma and suicidality.


Separation anxiety is an important clinical dimension, often with roots in childhood but likely to wax and wane across the lifespan and even to manifest for the first time during adulthood.50–52 The present study indicates that adult separation anxiety disorder and dimensional symptoms of separation anxiety, both in childhood and in adulthood, are associated with suicidal risk in outpatients with mood and anxiety disorders.

Submitted: February 17, 2020; accepted September 15, 2020.
Published online: March 16, 2021.
Potential conflicts of interest: The authors report no financial or other relationship relevant to the subject of this article.
Funding/support: This work was in part supported by the German Research Foundation (DFG), CRC-TRR58 “Fear, Anxiety, Anxiety Disorder” (projects C02 and Z02 to KD) and the Fondazione Cassa di Risparmio di La Spezia (to Dr Schiele).
Role of the sponsor: The supporting sources had no role in the design, conduct, and reporting of the study.
Acknowledgment: The authors thank Gabriele Massimetti, Professor of Statistics at University of Pisa, for insights on methodology and advice on statistical analyses. Prof Massimetti has no conflicts of interest to declare.
Additional information: Drs Pini, Abelli, Schiele, Baldwin, Bandelow, and Domschke are members of the Anxiety Disorders Research Network, European College of Neuropsychopharmacology.

Clinical points

  • Separation anxiety is an important clinical dimension, often with roots in childhood but likely to manifest across the lifespan; however, it is frequently underdiagnosed.
  • This study found that adult patients with separation anxiety disorder have an increased risk of suicide.
  • Separation anxiety disorder may be a key factor in understanding the relationship between depression and suicidality. Clinicians should screen for separation anxiety disorder in adult patients, as they screen for other psychiatric disorders.

Editor’s Note: We encourage authors to submit papers for consideration as a part of our Focus on Suicide section. Please contact Philippe Courtet, MD, PhD, at

Volume: 82

Quick Links: Anxiety , Depression (MDD) , Suicide


Buy this Article as a PDF