Introduction
Panic disorder (PD) is typically a long-lasting condition, characterized by recurrent unexpected panic attacks, persistent worry about experiencing future attacks and their consequences and/or subsequent maladaptive changes in behavior. 1 It affects 1.6%–2.2% of the world population,Reference Quilty, Van Ameringen, Mancini, Oakman and Farvolden 2 , Reference Freire, Hallak, Crippa and Nardi 3 with a lifetime prevalence of 1.5%–5%Reference Bandelow and Michaelis 4 and a 12-month prevalence between 1.8% and 2.7%.Reference Kessler, Wai, Jin, Ruscio, Shear and Walters 5 , Reference Goodwin, Faravelli and Rosi 6 Previous investigations suggest that the longer PD remains untreated, the greater the likelihood of nonresponse.Reference Freire, Hallak, Crippa and Nardi 3 Therefore, the duration of untreated illness (DUI), defined as the interval between the onset of a specific psychiatric disorder and the administration of the first appropriate pharmacological treatment according to guidelines in compliant subjectsReference Compton, West and Olfson 7 -Reference Dell’Osso, Camuri, Benatti, Buoli and Altamura 9 may be a potential predictor of clinical course and outcome.Reference Slaap and Den Boer 10 , Reference Altamura, Santini, Salvadori and Mundo 11
To date, DUI has been investigated across different psychiatric conditions, including major depressive disorder (MDD),Reference Altamura, Dell’Osso, Mundo and Dell’Osso 12 , Reference Altamura, Dell’Osso, D’Urso, Russo, Fumagalli and Mundo 13 -Reference Galimberti, Bosi, Volontè, Giordano, Dell’Osso and Viganò 15 bipolar disorder,Reference Buoli, Cesana and Fagiolini 16 obsessive–compulsive disorder (OCD),Reference Dell’Osso, Buoli, Hollander and Altamura 17 -Reference Matsumoto, Nakamae, Abe, Watanabe and Narumoto 19 somatization,Reference Herzog, Shedden-Mora, Jordan and Löwe 20 eating,Reference Neubauer, Weigel and Daubmann 21 and schizophrenia spectrum disorders.Reference Dell’Osso and Altamura 8 , Reference Dell’Osso, Glick, Baldwin and Altamura 22 -Reference Dell’Osso, Cremaschi and Palazzo 24 In respect to anxiety disorders, previous reports indicate a significantly shorter DUI compared to findings from studies conducted with patients with obsessive–compulsive, mood, and schizophrenia spectrum disorders.Reference Dell’Osso, Camuri, Benatti, Buoli and Altamura 9 , Reference Dell’Osso, Oldani and Camuri 25 , Reference Vigne, Fortes and Dias 26 Moreover, differences in terms of latency to treatment emerged between patients with PD and generalized anxiety disorder, with PD patients having a shorter DUI.Reference Dell’Osso, Camuri, Benatti, Buoli and Altamura 9 , Reference Benatti, Camuri and Dell’Osso 27 Indeed, patients with PD may wait for years before receiving an adequate pharmacological treatment, with possible negative prognostic implications, such as a higher risk of developing comorbid MDD.Reference Altamura, Santini, Salvadori and Mundo 11 , Reference Altamura, Dell’Osso, D’Urso, Russo, Fumagalli and Mundo 13
Nonetheless, only a few studies have assessed DUI in PD. A preliminary naturalistic study in a sample of 96 subjects attending an Italian outpatient clinic found a mean DUI of almost 4 years,Reference Altamura, Buoli, Albano and Dell’Osso 28 while a more recent investigation involving a larger sample (n = 138) showed a DUI of slightly more than 3 years.Reference Dell’Osso, Camuri, Benatti, Buoli and Altamura 9 Finally, 4–5 years DUI was reported in an Italian multicenter study conducted on a sample of 49 patients with PD.Reference Benatti, Camuri and Dell’Osso 27 Previous research has some methodological limitations, such as relatively small sample sizes, the lack of assessment of potentially important variables related to psychopathological onset and latency to treatment, including type of first referral, and presence of comorbidities and clinical presentation.
Focusing on pharmacological treatments of anxiety disorders and PD in particular, the widespread prescription of benzodiazepines (BDZ) as monotherapy might delay the initiation of guideline-recommended long-term treatments.Reference Bruce, Vasile and Goisman 29 Treatment guidelines support the efficacy of selective serotonin reuptake inhibitors (SSRIs) or the serotonin-norepinephrine reuptake inhibitor (SNRI) venlafaxine as first line pharmacological treatment of PD, and recommend caution for BDZ prescription due to associated risks of abuse, tolerance, and withdrawal.Reference Stein, Marcia Goin and Pollack 30 , 31 As little is known about the assessment and comparison between DUI-SSRI and DUI-BDZ in patients affected by PD, we sought to determine the mean DUI and related variables in a large sample of PD patients, with the additional aim of comparing findings with those from previous investigations in the field.
Methods
Study sample
A total of 314 patients with PD, who attended the outpatient psychotherapy unit of the Psychiatric Department of the Ospedale L. Sacco of Milan, Italy, between January 2015 and March 2019, was recruited. Data were retrospectively collected from patients’ medical records. Patients had been diagnosed according to The Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5) criteria by means of the Structured Clinical Interview for DSM-5 (SCID)Reference First, Spitzer, Gibbon and Williams 32 at the time of the first contact with the service. Subjects primarily affected by schizophrenia-spectrum disorders, unipolar or bipolar mood disorders, somatization, obsessive–compulsive and post-traumatic spectrum disorders, mild/major neurocognitive disorders, and anxiety disorders secondary to substance use or other medical conditions were excluded from the study. Comorbidity with personality disorders was allowed and assessed via the Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II),Reference Lobbestael, Leurgans and Arntz 33 administered by psychologists with specific training. Patients were interviewed after providing written informed consent for participation in the study and for having their clinical records examined for research purposes.
Assessment
Demographic and clinical variables collected during clinical interviews with patients (and, when available, relatives) were retrieved from medical records, including gender, age, education, age at onset, medical comorbidities, lifetime substance and alcohol abuse, age at first contact with psychiatric services, type of service at first contact (ie, emergency department, psychotherapy unit, child and adolescent psychiatry unit [CAPU], general practitioner [GP], psychiatrist, or other clinician), age at first pharmacological treatment, current pharmacological treatment, family history of psychiatric disorders and presence of agoraphobia or nocturnal panic attacks, and prevalent type of panic attacks. As previously proposed,Reference Compton, West and Olfson 7 , Reference Dell’Osso and Altamura 8 , Reference Dell’Osso, Benatti and Buoli 34 DUI-SSRI was defined as the interval (in months) between the onset of the disorder and the administration of the first adequate pharmacological treatment (ie, SSRI), at an appropriate dosage and for an adequate period of time, in compliant patients, in agreement with international treatment guidelines.Reference Bandelow, Baldwin and Zwanzger 35 , Reference Bandelow, Scott and Wedekind 36 Furthermore, DUI-BDZ, indicating the interval (in months) between PD onset and administration of benzodiazepines, was also calculated.
Statistical analysis
Descriptive analyses were carried out for sociodemographic and clinical variables of the total sample. Because of the non-Gaussian data distribution (assessed with the Shapiro–Wilk test), a nonparametric test (Wilcoxon matched-pair signed rank test) was applied to compare the distributions of DUI-SSRI vs DUI-BDZ. Welch’s t test was performed to compare results with findings from previous studies conducted by our group in independent samples. The level of significance was set at 0.05. Statistical analyses were performed using SPSS for Windows software (version 20; SPSS Inc., Chicago, IL) and Prism (GraphPad Prism version 9.0).
Results
The main sociodemographic and clinical variables of the sample are summarized in Table 1. Male to female ratio was about 1:2, with a mean age of 42.55 ± 16.11 years. Of the 314 participants, 202 (64.3%) had a positive psychiatric family history, specifically for MDD in the 19.4% of cases, anxiety disorders (17.8%), panic attacks (10.8%), alcohol use disorder (1.6%), schizophrenia (1.6%), bipolar disorder (1.3%), eating disorder (0.6%), and impulse control disorders (0.3%). As far as clinical variables were concerned, the mean age at onset was 31.74 ± 15.79 years, while the mean age at first referral was 35.28 ± 15.73 years. Most patients referred to either a psychiatrist (35.7%), a CAPU (2.9%), or a psychotherapy unit (22.3%), even though contact with other health services (emergency department, GP or other specialists) was also common (38.9%). More than half of the subjects suffered from comorbid psychiatric conditions, namely, in decreasing order: personality disorders (23.2%), other anxiety disorders (4.1%), MDD (3.8%), alcohol or susbtance use disorder (3.2%), post-traumatic stress disorder (PTSD, 1.3%), eating disorder (1%), somatization disorders (0.6%), bipolar disorder (0.3%), adjustment disorder (0.3%), OCD (0.3%), and gambling disorder (0.3%). In detail, with regard to personality disorders, 14% suffered from a cluster C disorder, 8.9% from a cluster B, and 1.4% from a cluster A disorder, whereas 9.6% suffered from multiple and 0.6% from unspecified personality disorders. Lifetime alcohol or substance use disorders were reported by 6.4% of the participants.
Sociodemographic and Clinical Variables of the Total Sample
Values for categorical and continuous variables are expressed as mean ± SD.
Abbreviations: BDZ, benzodiazepines; CAPU, child and adolescent psychiatric unit; DUI, duration of untreated illness; ER, emergency room; GP, general practitioner; MDD, major depressive disorder; NOS, not otherwise specified; PTSD, post-traumatic stress disorder; SSRI, selective serotonin reuptake inhibitors.
In terms of clinical presentation, almost two-thirds (71.3%) of patients predominantly experienced unexpected panic attacks. Nocturnal panic attacks were prevalent in a small fraction of the subjects (10.8%), whereas agoraphobia was reported by 61.5% of the participants. With respect to pharmacological treatment, the vast majority of patients had been prescribed with BDZs, either alone (31.2%) or in combination with SSRIs (29.3%), 11.2% only received treatment with antidepressants (either SSRI [9.6%] or tricyclic antidepressant [TCA, 0.3%] monotherapy or the combination of both [1.3%]), whereas the remainder (28.3%) were not taking any medication.
The mean DUI-SSRI of the total sample was 64.25 ± 112.74 months, while the mean DUI-BDZ was significantly shorter (35.09 ± 78.62; W = 6216, P < .0001; Figure 1). In addition, with regard to the DUI-SSRI, there were significant differences in comparison with previous results from Altamura et alReference Altamura, Buoli, Albano and Dell’Osso 28: mean DUI = 44.35 months (standard deviation [SD] ± 59.86, N = 96) and Dell’Osso et alReference Dell’Osso, Camuri, Benatti, Buoli and Altamura 9: mean DUI = 39.55 months (SD ± 57.25, N = 138), while not differing significantly from findings reported by Benatti et alReference Benatti, Camuri and Dell’Osso 27: mean DUI = 53.9 months (SD ± 81.5, N = 49; respectively: t = 2.256, P = .024; t = 3.082, P = .002; t = 0.780, P = .438; Figure 2).
Latency to different pharmacological treatments (DUI-BDZ vs DUI-SSRI) in patients with panic disorder; P < .05. Abbreviations: BDZ, benzodiazepines; DUI, duration of untreated illness; SSRI: selective serotonin reuptake inhibitors.
Differences in terms of DUI-SSRI (expressed in months) in patients with Panic Disorder between the present study and previous reports investigating latency to pharmacological treatments; P < .05. Abbreviations: DUI, duration of untreated illness; SSRI, selective serotonin reuptake inhibitors.
Discussion
We sought to explore and compare latency to different pharmacological treatments (SSRIs vs BDZs) in a sample of 314 Italian outpatients affected by PD.
In accordance with the literature,Reference Bandelow and Michaelis 4 , Reference Katzman, Bleau and Blier 37 patients with PD were mostly women, showing a male to female ratio of almost 1:2. In addition, around two-thirds of participants had a family history of psychiatric disorder, with an almost 15% rate of familiality for PD or panic attacks. With respect to age at onset, the mean value observed in the sample was approximately 32 years, which is consistent with previous findingsReference Dell’Osso, Camuri, Benatti, Buoli and Altamura 9 , Reference Altamura, Buoli, Albano and Dell’Osso 28 confirming a common onset of PD in early adulthood. Interestingly, age at first referral was slightly older (around 35 years). This result is also consistent with previous reportsReference Dell’Osso, Camuri, Benatti, Buoli and Altamura 9 and highlights how panic symptoms are still underrecognized and, consequently, under-treated. More than one-third of patients had contacted healthcare services outside the mental health field (emergency room, GP or other specialists), which might contribute to a delay in proper diagnosis and treatment. Moreover, PD is frequently comorbid with medical illnesses, whose symptoms often overlap and complicate its diagnostic recognition.Reference Meuret, Kroll and Ritz 38 Our results confirmed this association, with one-fifth of patients presenting physical comorbidities. Although a frequent association with MDD, dysthymia, BD, other anxiety disorders, and personality disorders has been reported, with at least one lifetime comorbid condition in up to 80% of the patients with PD,Reference Bandelow and Michaelis 4 , Reference Kessler, Wai, Jin, Ruscio, Shear and Walters 5 , Reference de Jonge, Roest and Lim 39 in our sample, around 50% of patients had a comorbid psychiatric disorder, MDD, BD, and other anxiety disorders being less prevalent than previously reported. In terms of prevalence of personality disorders, 35.7% of our patients met DSM-IV criteria for at least one personality disorder, a result that slightly differs from earlier studies reporting possible comorbidity rates between 40% and 65%.Reference Latas, Starcevic, Trajkovic and Bogojevic 40 , Reference Latas, Vučinić Latas and Spasić Stojaković 41 , Reference Ozkan and Altindag 42 The most represented personality disorders belonged to the C cluster (14%), confirming the well established link between PD and anxious personality traits.Reference Telch, Kamphuis and Schmidt 43 , Reference Milrod, Leon, Barber, Markowitz and Graf 44 Coexistence of multiple personality disorders was not uncommon (almost 10% of the sample). Although co-occurrence of panic and alcohol and substance use disorders (SUD) is of frequent observation in patients with PD,Reference Lopez, Turner and Saavedra 45 -Reference Swendsen, Merikangas, Canino, Kessler, Rubio-Stipec and Angst 49 with an estimated prevalence of any SUD above 20%,Reference Kessler, Wai, Jin, Ruscio, Shear and Walters 5 our findings showed lower lifetime prevalence rates, with only 6.4% of participants meeting DSM-5 criteria for SUD. This result needs to be cautiously interpreted and might be partly explained by potential reporting bias and the frequent unavailability of relatives or external referents and records from outpatient substance abuse services, with possible underestimation of such comorbidity pattern.
More than two-thirds of participants reported the predominance of unexpected panic attacks, although only around 10% experienced nocturnal attacks regularly, a slightly lower value than previously documented.Reference Levitan and Nardi 50 A possible explanation may be that all participants were receiving cognitive-behavioral psychotherapy (CBT) and some were taking medications that might have reduced their occurrence. In line with previous observations,Reference Ozkan and Altindag 42 agoraphobia was diagnosed in approximately 60% of participants.
As far as treatment is concerned, the gold standard first-line treatment for PD according to major international guidelines comprises either SSRIs or the SNRI venlafaxine and/or nonpharmacological interventions such as CBT.Reference Stein, Marcia Goin and Pollack 30 , 31 , Reference Katzman, Bleau and Blier 37 , Reference Baldwin, Anderson and Nutt 51 TCAs have also been found efficacious in PD, but are not recommended as first-line treatment due to safety and tolerability issues.Reference Bandelow, Baldwin and Zwanzger 35 BDZs may be useful for acute anxiety or agitation or while waiting for the onset of the efficacy of antidepressantsReference Stein, Marcia Goin and Pollack 30 , 31 , Reference Katzman, Bleau and Blier 37 , Reference Baldwin, Anderson and Nutt 51: longer-term use of these compounds, however, is not recommended because of the potential risk of abuse, sedation, cognitive impairment, and psychomotor alterations,Reference Baldwin, Anderson and Nutt 51 , Reference Ströhle, Gensichen and Domschke 52 even though some research supports their use in comparison to antidepressants,Reference Offidani, Guidi, Tomba and Fava 53 likely in patients with specific characteristics. In fact, BDZs are not effective for the treatment of conditions that are often comorbid with PD, such as MDD or OCD.Reference Katzman, Bleau and Blier 37 In the present sample, we analyzed the current pharmacological treatment rate and found that approximately one-third of patients was not taking any medication, while more than two-thirds were receiving BDZ treatment, either as monotherapy or in conjunction with SSRIs. Only around 10% of the sample was taking antidepressants alone. These results highlight the discrepancy between recommendations from international treatment guidelines and routine clinical practice, and are consistent with previous findings reported in the literature.Reference Kessler, Wai, Jin, Ruscio, Shear and Walters 5
We documented a mean latency to first guideline-recommend PD treatment of more than 5 years. As already mentioned, a growing body of literature has investigated DUI in patients affected by PD,Reference Dell’Osso, Camuri, Benatti, Buoli and Altamura 9 , Reference Dell’Osso, Oldani and Camuri 25 , Reference Vigne, Fortes and Dias 26 , Reference Benatti, Camuri and Dell’Osso 27 , Reference Altamura, Buoli, Albano and Dell’Osso 28 reporting values ranging between 3 and 5 years. The acute onset of PD, its symptoms having a great impact on individual functioning, the presence of physical symptomatology and different degrees of insight might be expected to bring affected individuals to seek professional help earlier than patients with other mental disorders—around 60% of patients with PD reported contact with health care services because of panic symptoms at least onceReference Goodwin, Faravelli and Rosi 6 and around 30% have contact within 1 year of the onset.Reference Iza, Olfson, Vermes, Hoffer, Wang and Blanco 54 Nevertheless, the condition remains often under-recognized, as our report seems to confirm. Although the majority of individuals with PD without agoraphobia, and nearly 90% of those with PD and agoraphobia are recognized by their GPs as having a psychiatric disorder,Reference Goodwin, Faravelli and Rosi 6 only a minority receives proper treatments, with a negative impact on outcome and response to treatment.Reference Freire, Hallak, Crippa and Nardi 3 , Reference Altamura, Santini, Salvadori and Mundo 11 , Reference Altamura, Camuri and Dell’Osso 55 In this respect, we detected a mean latency to the first BDZ prescription of approximately 3 years. In addition, when we compared DUI-SSRI vs DUI-BDZ, we observed a statistically significant difference (of approximately 2.5 years) between these two measurements. This finding indicates that, despite the guideline recommendations, BDZs continue to be widely prescribed as first treatment in PD,Reference Bruce, Vasile and Goisman 29 even though existing evidence suggests that long-term BDZ prescription might delay proper treatment in patients with mood and anxiety disorders, including PD.Reference De Carlo, Grancini and Vismara 56 , Reference Grancini, De Carlo and Palazzo 57
Finally, we compared latency to SSRI treatment with findings from previous studies and observed some significant differences. The longer DUI in the present study could reflect a longer latency to treatment in patients with PD, but may be influenced by the setting of our study: all participants were undergoing CBT treatment, and may have been more inclined to access a nonpharmacological intervention. Nonetheless, this result further confirms the substantial delay in the administration of an adequate pharmacological treatment in patients with PD, stressing the need for early diagnosis and treatment in order to achieve a better clinical outcome.
In the interpretation of the results, the following limitations need to be considered. First, the naturalistic way of collecting data relied on recall by patients and available family members, who may not have always been precise. Second, our findings are related to treatment-seeking patients so may not be representative of the wider population of individuals with PD. Third, local mental health service delivery factors and cultural attitudes may have influenced patients’ access, diagnosis, and treatment. Finally, all participants were receiving CBT, which is considered among first‐line treatment options for PD according to international guidelines.Reference Stein, Marcia Goin and Pollack 30 , 31 , Reference Katzman, Bleau and Blier 37 , Reference Baldwin, Anderson and Nutt 51 In this perspective, it would be interesting to compare latency to CBT and/or other psychotherapy treatments versus latency to pharmacological treatment, and to examine their potential interactions.
Conclusions
The present study reported a significant difference between DUI-BDZ and DUI-SSRI in patients with PD and a longer DUI-SSRI than previously observed. Such results might be of relevance with regard to clinical outcome and overall prognosis in patients with PD. Further and longitudinal analyses in larger samples are needed to confirm or refute reported findings and provide additional insight into the factors influencing DUI and its relationship with clinical course, outcome and morbidity of PD.
Acknowledgments
All authors were involved at all stages of manuscript development, approved the final version of the manuscript to be published, and agreed to act as guarantors of the work. B.D., D.B., and K.D. are members of the European College of Neuropsychopharmacology (ECNP) Anxiety Disorders Research Network (ADRN) Thematic Working Group.
Funding
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Disclosures
Prof. Dell’Osso has received Grant/Research Support from LivaNova, Inc., Angelini, and Lundbeck and Lecture Honoraria from Angelini, Lundbeck, Janssen, Neuraxpharma, and Pfizer. Prof. Domschke is a member of the Neuroscience Steering Committee, Janssen Inc. Prof. Baldwin is a Medical Patron of Anxiety UK and President-Elect of the British Association for Psychopharmacology. Eleonora Piccoli, Irma Bergamaschini, Laura Molteni, Simone Vanzetto, Alberto Varinelli, Caterina Viganò, Gabriele Catania, David Baldwin, and Katharina Domschke report no conflict of interest nor anything to disclose in relation to the content of the present article.
Author Contributions
Conceptualization: B.D., I.B., and A.V.; Project administration: B.D.; Supervision: B.D. and C.A.V.; Resources: G.C.; Data curation: I.B., A.V., and E.P.; Investigation: I.B.; Formal analysis: E.P.; Methodology: E.P.; Visualization: E.P.; Writing – original draft: L.M., S.V., and E.P.; Writing – review & editing: B.D., D.S.B., K.D., and E.P.
