A Mechanism-Based, Evidence-Graded Framework for Managing Panic Attacks and Panic Disorder

Abstract

Background. Panic attacks are common, and panic disorder is a disabling, treatable condition. Public and clinical guidance on managing attacks often rests on tradition (relaxation, deep breathing, distraction) rather than on dismantling research that identifies which treatment components work.

Objective. To synthesize evidence from component network meta-analyses, delivery-format meta-analyses, randomized dismantling trials, experimental work on safety behaviors, and guidelines into a stratified framework linking mechanism to intervention.

Findings. Cognitive models hold that panic is maintained by catastrophic misinterpretation of bodily sensations and by avoidance and safety behaviors that block disconfirmation. Consistent with this, interoceptive exposure is the component most strongly associated with efficacy in component network meta-analysis, while applied muscle relaxation is associated with lower efficacy and breathing retraining shows no clear incremental benefit. CBT is the best-supported psychological treatment; guided delivery (clinician-led or guided digital) outperforms unguided self-help. SSRIs/SNRIs are first-line pharmacotherapy; benzodiazepines are not recommended as first-line treatment. Sensory grounding and slow breathing are plausible in-the-moment aids but lack panic-specific trial evidence and may function as safety behaviors if used to avoid sensations.

Conclusions. We propose a three-tier model (acute-episode response, exposure-based consolidation, relapse prevention) with explicit evidence grades, and identify priority gaps: direct comparisons of in-the-moment techniques, long-term outcomes of component-optimized CBT, and implementation of guided digital care.

Keywords: panic disorder; panic attacks; interoceptive exposure; cognitive behavioral therapy; safety behaviors; anxiety sensitivity; inhibitory learning; component network meta-analysis

Key points

• Panic is maintained by interpretation and avoidance, not by arousal alone; the therapeutic target is the response to sensations.

• Interoceptive exposure has the strongest component-level evidence; relaxation-based components have the weakest.

• During an attack, the goal is staying with and riding out the surge, not eliminating it.

• Guided CBT (in person or digital) is first-line; unguided self-help is weaker.

• Many popular techniques are evidence-poor for panic specifically; they should be framed as optional, non-compulsory aids.

1. Introduction

A panic attack is an abrupt surge of intense fear or discomfort peaking within minutes, accompanied by symptoms such as palpitations, dyspnea, chest discomfort, dizziness, derealization, and fear of dying or losing control (American Psychiatric Association, 2022). Isolated attacks are common in the general population; panic disorder is diagnosed when attacks are recurrent and unexpected and are followed by persistent worry about further attacks or their consequences, or by maladaptive behavior change. Lifetime prevalence of panic disorder in the United States is roughly 5% (Kessler et al., 2006), and the condition is associated with substantial functional impairment, health-care use, and comorbidity (Roy-Byrne et al., 2006; Craske & Stein, 2016).

Effective treatments exist, yet advice reaching patients and the public frequently emphasizes relaxation and controlled breathing. The question this review addresses is practical: which strategies, delivered in which order and format, are supported by the best available evidence, and what mechanisms explain why?

This article contributes (a) an integrated mechanism-to-intervention framework, (b) explicit evidence grading that separates well-supported components from plausible but weakly tested techniques, and (c) a stepped-care pathway that includes safety considerations often omitted from self-help guidance.

2. Methods (narrative review)

This is a narrative critical review. Sources were prioritized by evidence hierarchy: component and conventional network meta-analyses and Cochrane reviews; randomized dismantling trials; experimental studies of mechanisms; and international guidelines (WFSBP, NICE, BAP). [Authors: insert your actual search dates, databases, search strings, inclusion rules, and a flow description. If a systematic claim is intended, follow PRISMA 2020 and register the protocol; otherwise retain the "narrative review" label and follow the SANRA quality criteria.] Evidence grades in Table 1 are the authors' qualitative judgments, not formal GRADE ratings.

3. Mechanisms: why panic persists

3.1 The cognitive-behavioral maintenance loop

Cognitive models propose that panic arises when benign bodily sensations are interpreted as imminent catastrophe (e.g., cardiac arrest, suffocation, "going insane"), producing apprehension, further autonomic arousal, and apparent confirmation of the threat (Clark, 1986). Earlier formulations described this as "fear of fear" (Goldstein & Chambless, 1978). The vulnerability factor of anxiety sensitivity, the belief that anxiety-related sensations are harmful, predicts panic and response to treatment (Reiss et al., 1986).

3.2 Safety behaviors and avoidance

Two maintenance processes are central. Avoidance (of exercise, caffeine, crowds, hot rooms) preserves feared appraisals by preventing exposure to sensations. Safety behaviors (carrying medication, sitting near exits, rigidly controlling breathing, constant pulse checking) lead people to attribute the absence of catastrophe to the behavior rather than to the benign nature of the sensation (Salkovskis, 1991). Experimental work indicates that safety-seeking during exposure can reduce fear reduction and impair belief change (Sloan & Telch, 2002). This matters for any technique, however comforting, that is used as a compulsory "rescue."

3.3 Inhibitory learning

Modern exposure theory holds that treatment does not erase the fear association but builds a competing, inhibitory "safe" association that must outcompete it across contexts and over time (Craske et al., 2008, 2014). Practical implications include maximizing expectancy violation (testing the feared prediction), varying contexts and intensity, and removing safety signals. This framework is the logical basis of interoceptive exposure.

3.4 Physiology: what the sensations are and are not

Panic involves a surge of sympathetic activation (tachycardia, hyperventilation, sweating, trembling). These sensations are aversive but not themselves indicative of cardiac or respiratory danger in a person without underlying disease. Psychoeducation about this should be accurate and not absolute: the claim is not that symptoms can never signal medical illness (see Section 8) but that, once serious causes are reasonably excluded, panic sensations are not harmful in themselves. Fainting is uncommon in panic because blood pressure typically rises rather than falls, in contrast to blood-injury-injection phobia.

4. Evidence base for psychological treatment

4.1 CBT overall

Meta-analyses place CBT for panic disorder in the large effect range against inactive controls (Hedges g ≈ 0.8; Cuijpers et al., 2016; see also Hofmann, 2012). In a network meta-analysis of psychotherapies for panic disorder, CBT was the only psychotherapy that remained superior to treatment as usual after removing trials at high risk of bias (SMD −0.67, 95% CI −0.95 to −0.39; Papola et al., 2022). Long-term follow-up studies show that gains are largely maintained (e.g., Fava et al., 2001). Guidelines uniformly rank CBT among first-line treatments (Bandelow et al., 2023; NICE, 2011; Baldwin et al., 2014).

4.2 Component-level evidence

CBT is a package. A component network meta-analysis of 72 trials found that interoceptive exposure was the component most strongly associated with greater efficacy and acceptability, whereas muscle relaxation was associated with lower efficacy; breathing retraining contributed little to efficacy (Pompoli et al., 2018). The most- versus least-efficacious combinations differed substantially in remission odds, showing that "CBT" is not an interchangeable label. These findings are consistent with randomized dismantling studies: adding breathing retraining to CBT did not improve and in places worsened outcomes (Schmidt et al., 2000), and a comparison within CBT found interoceptive exposure favorable to breathing retraining on key outcomes at follow-up (Craske et al., 1997). A broader account of how respiratory training can help or hinder is given by Meuret et al. (2003). Capnometry-guided approaches that raise CO₂ tolerance rather than teach "calming" breathing are an active line of research (Meuret et al., 2018).

Interpretive caution. Component network meta-analyses are observational at the component level: components are not randomized independently, and confidence in several estimates is modest. They support prioritizing interoceptive exposure, not the claim that relaxation or breathing work is harmful.

4.3 Interoceptive exposure in practice

Interoceptive exposure involves deliberate, repeated induction of feared sensations (e.g., spinning for dizziness, voluntary hyperventilation for lightheadedness/tingling, straw breathing for air hunger, stair-running for racing heart) while dropping avoidance and safety behaviors, in a graded hierarchy built from a sensation-sensitivity symptom induction test. Delivery should follow inhibitory learning principles: state the feared prediction beforehand, rate expected and observed outcomes, repeat to violate expectancy, and vary context (Craske et al., 2014). Digital CBT meta-analysis also finds interoceptive exposure and personalization linked to better effects (Jung et al., 2025).

4.4 Cognitive restructuring

Cognitive restructuring (identifying catastrophic predictions, weighing evidence, generating alternatives) is a standard component, but its distinct contribution is less clear than that of exposure in component analyses. Its practical role is to frame experiments: behavioral tests make appraisals revisable at an experiential rather than purely verbal level.

4.5 Delivery formats

A network meta-analysis found that group CBT (SMD −0.47), individual CBT (−0.43), and guided self-help (−0.42) outperformed treatment as usual, whereas unguided self-help (−0.21) did not (Papola et al., 2023). A meta-analytic review of digital CBT reported an effect of g = 0.70 versus passive controls and no significant difference from face-to-face CBT (Jung et al., 2025; see also Andrews et al., 2018). Guidance, not merely digitization, appears to be the active ingredient for scalability.

5. Acute-episode management

The aim during an attack is not to abolish symptoms but to avoid the loop that amplifies them, and to avoid training the brain that the sensations required a rescue.

5.1 Reappraisal and acceptance. Remind oneself of an accurate, pre-learned formulation ("this is adrenaline, it peaks and passes, it is uncomfortable but not dangerous"), and let sensations be present while remaining in the situation where possible. This follows directly from the exposure rationale.

5.2 Sensory grounding (e.g., 5-4-3-2-1). Naming sensory objects directs attention outward and is widely taught in clinical settings. However, panic-specific randomized evidence is lacking. The frequently cited quantitative support comes from a pre–post study of test anxiety in nursing students, which is non-comparable (different construct, population, and likely no control condition) and should not be read as evidence for panic attacks. Mechanistically, grounding is distraction/attention redirection; distraction can interfere with inhibitory learning if used to avoid sensations (see Section 3.2). Recommended framing: a short-term option to reduce derealization-related distress, used flexibly and not as a ritual, combined with continued exposure practice.

5.3 Breathing. Slow breathing with a prolonged exhale modestly engages vagal pathways and may lower arousal, and slow-paced breathing with HRV biofeedback altered inflammatory markers in a small trial in panic disorder (Herhaus et al., 2023). Yet this does not establish clinical efficacy for acute panic, and dismantling evidence offers no support for adding breathing retraining to CBT (Schmidt et al., 2000; Pompoli et al., 2018). Position it as optional, never as a compulsory technique.

5.4 Medication as rescue. PRN benzodiazepine use can act as a safety behavior and carries dependence and cognitive risks; it should not be a default acute strategy (see Section 6).

5.5 After the attack. Review what was predicted and what happened; resume planned activities rather than retreating.

6. Pharmacotherapy

SSRIs and SNRIs are first-line pharmacotherapy (Bandelow et al., 2023; Baldwin et al., 2014; NICE, 2011). Starting doses should be low and titrated, because early jitteriness can mimic panic. Benzodiazepines are effective short term but are limited by dependence risk, cognitive effects, and worse long-term outcomes, and are generally not first-line (Bighelli et al., 2018; Offidani et al., 2013). Continuation for at least 6–12 months after remission before supervised, gradual tapering is commonly recommended (Bandelow et al., 2023). Combination treatment may help in selected patients, though evidence that combining adds benefit over CBT alone is mixed (Barlow et al., 2000). Medication decisions belong to a prescriber and should account for pregnancy, comorbidity, and patient preference.

7. Lifestyle and adjunctive factors

Aerobic exercise has the most consistent supportive evidence among lifestyle interventions for anxiety, with small-to-moderate effects (Stubbs et al., 2017) and reductions in anxiety sensitivity (Smits et al., 2008). Because exercise produces interoceptive sensations, it can serve as graded exposure when introduced gradually and without safety behaviors. Reviewing caffeine, nicotine, alcohol, cannabis, stimulants, and sleep deprivation is sensible, as each can provoke or maintain arousal. Evidence for most complementary approaches remains limited (Locke et al., 2015; Ströhle et al., 2018).

8. Safety, assessment, and special considerations

• Rule out organic mimics at first presentation: cardiac arrhythmia, ischemia, thyroid disease, pheochromocytoma, hypoglycemia, asthma/COPD, vestibular disorders, seizure disorders, and substance intoxication or withdrawal. New chest pain, exertional symptoms, syncope, or risk factors warrant medical assessment before an attack is assumed to be panic.

• Suicide risk and comorbidity (depression, substance use, PTSD) should be screened for; panic disorder is associated with elevated suicidal ideation in the context of comorbidity.

• Exposure suitability: screen for cardiovascular, respiratory, seizure, and pregnancy considerations before induction exercises; adapt (e.g., avoid hyperventilation in some medical conditions).

• Access and equity: guided digital CBT can reduce barriers but requires attention to digital literacy, language, and cultural fit.

9. A stepped framework

Stepping. Mild–moderate: guided self-help or guided digital CBT plus lifestyle review. Moderate–severe, or inadequate response: clinician-led CBT and/or SSRI/SNRI. Non-response: review diagnosis, adherence, comorbidity, and fidelity (is exposure actually happening?), then augment or switch.

Table 1. Technique evidence summary

10. Limitations of the evidence and of this review

Component network meta-analyses are subject to confounding by co-occurring components. Many trials are small, with variable control conditions and short follow-up. Sampling is predominantly from Western clinical populations. Publication bias and researcher allegiance are plausible. Evidence grading here is qualitative. This review is narrative and may miss relevant studies. Clinical translation of mechanistic experiments (e.g., on safety behaviors) is partly inferential.

11. Future directions

(1) Randomized, head-to-head tests of grounding, slow breathing, and acceptance-based acute strategies in panic disorder. (2) Component-optimized CBT trials with long follow-up. (3) Predictors of response to interoceptive exposure (anxiety sensitivity, respiratory physiology). (4) Implementation studies of guided digital CBT, including equity outcomes. (5) Personalization using biomarkers, with attention to avoiding overinterpretation of small mechanistic studies.

12. Conclusion

Panic attacks are best understood as a learned alarm response maintained by catastrophic interpretation and by avoidance. The strongest evidence supports exposure-based CBT, with interoceptive exposure as its most distinctive active component, and supports SSRIs/SNRIs as first-line medication. Popular relaxation-oriented techniques have modest or no evidence for panic specifically and risk becoming safety behaviors. The practical message for patients and clinicians is not to learn to relax away the sensations, but to let them come, test the feared prediction, and discover that the alarm is safe to ride out.

References

All entries should be checked against the original sources (volume, page, DOI) before submission; I could not verify every bibliographic detail in this environment.

American Psychiatric Association. (2022). Diagnostic and statistical manual of mental disorders (5th ed., text rev.). APA.

Andrews, G., Basu, A., Cuijpers, P., et al. (2018). Computer therapy for the anxiety and depression disorders is effective, acceptable and practical health care: An updated meta-analysis. Journal of Anxiety Disorders, 55, 70–78.

Baldwin, D. S., Anderson, I. M., Nutt, D. J., et al. (2014). Evidence-based pharmacological treatment of anxiety disorders, PTSD and OCD: A revision of the 2005 BAP guidelines. Journal of Psychopharmacology, 28(5), 403–439.

Bandelow, B., Allgulander, C., Baldwin, D. S., et al. (2023). WFSBP guidelines for treatment of anxiety, obsessive-compulsive and posttraumatic stress disorders – Version 3. Part I: Anxiety disorders. World Journal of Biological Psychiatry, 24(2), 79–117.

Barlow, D. H., Gorman, J. M., Shear, M. K., & Woods, S. W. (2000). Cognitive-behavioral therapy, imipramine, or their combination for panic disorder. JAMA, 283(19), 2529–2536.

Bighelli, I., Trespidi, C., Castellazzi, M., et al. (2018). Antidepressants and benzodiazepines for panic disorder in adults. Cochrane Database of Systematic Reviews, CD011567.

Clark, D. M. (1986). A cognitive approach to panic. Behaviour Research and Therapy, 24(4), 461–470.

Craske, M. G., Kircanski, K., Zelikowsky, M., et al. (2008). Optimizing inhibitory learning during exposure therapy. Behaviour Research and Therapy, 46(1), 5–27.

Craske, M. G., Treanor, M., Conway, C. C., Zbozinek, T., & Vervliet, B. (2014). Maximizing exposure therapy: An inhibitory learning approach. Behaviour Research and Therapy, 58, 10–23.

Craske, M. G., Rowe, M., Lewin, M., & Noriega-Dimitri, R. (1997). Interoceptive exposure versus breathing retraining within cognitive-behavioural therapy for panic disorder with agoraphobia. British Journal of Clinical Psychology, 36(1), 85–99.

Craske, M. G., & Stein, M. B. (2016). Anxiety. The Lancet, 388(10063), 3048–3059.

Cuijpers, P., Cristea, I. A., Karyotaki, E., Reijnders, M., & Huibers, M. J. H. (2016). How effective are cognitive behavior therapies for major depression and anxiety disorders? A meta-analytic update of the evidence. World Psychiatry, 15(3), 245–258.

Fava, G. A., Rafanelli, C., Grandi, S., et al. (2001). Long-term outcome of panic disorder with agoraphobia treated by exposure. Psychological Medicine, 31(5), 891–898.

Goldstein, A. J., & Chambless, D. L. (1978). A reanalysis of agoraphobia. Behavior Therapy, 9(1), 47–59.

Herhaus, B., Conrad, R., & Petrowski, K. (2023). Effect of a slow-paced breathing with heart rate variability biofeedback intervention on pro-inflammatory cytokines in individuals with panic disorder: A randomized controlled trial. Journal of Affective Disorders, 326, 132–138.

Hofmann, S. G., Asnaani, A., Vonk, I. J. J., Sawyer, A. T., & Fang, A. (2012). The efficacy of cognitive behavioral therapy: A review of meta-analyses. Cognitive Therapy and Research, 36(5), 427–440.

Jung, H. W., Jang, K. W., Nam, S., et al. (2025). Digital cognitive behavioral therapy for panic disorder and agoraphobia: A meta-analytic review of clinical components to maximize efficacy. Journal of Clinical Medicine, 14(5), 1771.

Kessler, R. C., Chiu, W. T., Jin, R., et al. (2006). The epidemiology of panic attacks, panic disorder, and agoraphobia in the National Comorbidity Survey Replication. Archives of General Psychiatry, 63(4), 415–424.

Locke, A. B., Kirst, N., & Shultz, C. G. (2015). Diagnosis and management of generalized anxiety disorder and panic disorder in adults. American Family Physician, 91(9), 617–624.

Meuret, A. E., Rosenfield, D., Seidel, A., Bhaskara, L., & Hofmann, S. G. (2010). Respiratory and cognitive mediators of treatment change in panic disorder. Journal of Consulting and Clinical Psychology, 78(5), 691–704. (include only if cited in text)

Meuret, A. E., Wilhelm, F. H., Ritz, T., & Roth, W. T. (2003). Breathing training for treating panic disorder: Useful intervention or impediment? Behavior Modification, 27(5), 731–754.

Meuret, A. E., et al. (2018). Hypoventilation therapy alleviates panic by repeated induction of dyspnea. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 3(6), 539–545.

National Institute for Health and Care Excellence. (2011). Generalised anxiety disorder and panic disorder in adults: Management (CG113).

Offidani, E., Guidi, J., Tomba, E., & Fava, G. A. (2013). Efficacy and tolerability of benzodiazepines versus antidepressants in anxiety disorders: A systematic review and meta-analysis. Psychotherapy and Psychosomatics, 82(6), 355–362.

Papola, D., Ostuzzi, G., Tedeschi, F., et al. (2022). Comparative efficacy and acceptability of psychotherapies for panic disorder with or without agoraphobia: Systematic review and network meta-analysis of randomised controlled trials. British Journal of Psychiatry, 221(3), 507–519.

Papola, D., et al. (2023). CBT treatment delivery formats for panic disorder: A systematic review and network meta-analysis of randomised controlled trials. Psychological Medicine, 53(3), 614–624.

Pompoli, A., Furukawa, T. A., Imai, H., et al. (2018). Dismantling cognitive-behaviour therapy for panic disorder: A systematic review and component network meta-analysis. Psychological Medicine, 48(12), 1945–1953.

Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986). Anxiety sensitivity, anxiety frequency and the prediction of fearfulness. Behaviour Research and Therapy, 24(1), 1–8.

Roy-Byrne, P. P., Craske, M. G., & Stein, M. B. (2006). Panic disorder. The Lancet, 368(9540), 1023–1032.

Salkovskis, P. M. (1991). The importance of behaviour in the maintenance of anxiety and panic: A cognitive account. Behavioural Psychotherapy, 19(1), 6–19.

Schmidt, N. B., Woolaway-Bickel, K., Trakowski, J., et al. (2000). Dismantling cognitive-behavioral treatment for panic disorder: Questioning the utility of breathing retraining. Journal of Consulting and Clinical Psychology, 68(3), 417–424.

Sloan, T., & Telch, M. J. (2002). The effects of safety-seeking behavior and guided threat reappraisal on fear reduction during exposure: An experimental investigation. Behaviour Research and Therapy, 40(3), 235–251.

Smits, J. A. J., Berry, A. C., Rosenfield, D., et al. (2008). Reducing anxiety sensitivity with exercise. Depression and Anxiety, 25(8), 689–699.

Stubbs, B., Vancampfort, D., Rosenbaum, S., et al. (2017). An examination of the anxiolytic effects of exercise for people with anxiety and stress-related disorders: A meta-analysis. Psychiatry Research, 249, 102–108.

Ströhle, A., Gensichen, J., & Domschke, K. (2018). The diagnosis and treatment of anxiety disorders. Deutsches Ärzteblatt International, 155(37), 611–620.

[Figure could not be pasted. Copy it as a PNG image and paste it here.][Figure could not be pasted. Copy it as a PNG image and paste it here.][Figure could not be pasted. Copy it as a PNG image and paste it here.]