Authors: Stephen B. Lo (1Department of Psychology, The Ohio State University, Columbus, OH), Nicole A. Arrato (1Department of Psychology, The Ohio State University, Columbus, OH), Carolyn J. Presley (2Department of Internal Medicine, Division of Medical Oncology, The Ohio State University Comprehensive Cancer Center, Columbus, OH), Heather L. McGinty (3Department of Psychiatry, The Ohio State University Wexner Medical Center, Columbus, OH), Michael W. Otto (4Department of Psychological and Brain Sciences, Boston University, Boston, MA), Barbara L. Andersen (1Department of Psychology, The Ohio State University, Columbus, OH)
Categories: Article, dyspnea, exposure, anxiety sensitivity, lung cancer, cognitive-behavioral
Source: Annals of the American Thoracic Society
Authors: Stephen B. Lo, Nicole A. Arrato, Carolyn J. Presley, Heather L. McGinty, Michael W. Otto, Barbara L. Andersen
Dyspnea (breathlessness) commonly impacts patients with lung cancer, worsening depression, anxiety, quality of life, and functioning. Current treatments are limited.
To test the acceptability, feasibility, and preliminary efficacy of “Take a Breath” (TAB), a novel cognitive-behavioral treatment for dyspnea.
A randomized controlled trial compared TAB with standard of care (SOC) in patients with lung cancer reporting at least moderate dyspnea (N=45). TAB consisted of 5 one-hour, weekly, individual sessions employing exposure-based interventions paired with pulse oximetry biofeedback, psychoeducation, and behavioral skills (e.g., pursed lip breathing). Robust mixed-effects modeling tested group × time interactions.
The Client Satisfaction Questionnaire-8 measured acceptability. Accrual, treatment retention, and homework completion measured feasibility. Primary outcomes were the ATS Dyspnea Scale (dyspnea-related functioning) and Cancer Dyspnea Scale (dyspnea-related effort, discomfort, and anxiety). Secondary outcomes included depression (PHQ-9), health-related quality of life (SF-12), physical activity (IPAQ-SF), and functional status (KPS). Measurements occurred at baseline, mid-treatment (3 weeks), post-treatment (6 weeks), and 1-month follow-up.
TAB was ≥”mostly satisfactory” for 75% of participants. The accrual was 25.6%, with 60% completing all sessions and an 88.7% homework completion rate. Intent-to-treat analysis revealed greater improvements in TAB than SOC for dyspnea-related functioning (Cohen’s d=0.82, p=0.03) and anxiety (Cohen’s d=0.87, p<0.01) at post-treatment and follow-up. TAB outperformed SOC in improving depressive symptoms, health-related quality of life, sedentary time, and performance status over time (*p’*s<0.05).
TAB yielded symptom, psychological, and functional improvements, establishing its readiness for further testing as the first comprehensive cognitive-behavioral treatment for dyspnea and related sequelae.
Dyspnea, the subjective experience of breathing discomfort related to breathlessness (2), is reliably associated with functional impairment and worsened quality of life and mental health outcomes in patients with lung cancer (3). Despite the profound burden of dyspnea prevalent in up to 80% of patients with lung cancer at diagnosis (4, 5), current treatments for dyspnea are largely ineffective. Opioids—the most studied treatment for dyspnea—have limited efficacy and carry significant risks (6, 7). Anxiolytics demonstrate no reliable efficacy and supplemental oxygen is only beneficial in patients with hypoxemia (6). Dyspnea persists or worsens in two-thirds of patients with lung cancer receiving chemotherapy (8) and newer cancer-directed therapeutics (e.g., immunotherapy) having no relative benefit for dyspnea (9, 10). Medical interventions, particularly those targeting the cause of dyspnea, are crucial for effective care (e.g., thoracentesis). However, the limited success of current approaches may stem from insufficiently addressing the multifaceted factors that contribute to dyspnea, including physiological, psychological, social, and environmental (11). The 2021 American Society of Clinical Oncology (ASCO) guidelines for dyspnea recommend behavioral strategies (e.g., paced breathing, posture relief, relaxation) as they are low-risk and complement medical management, but limited efficacy data have hindered uptake (3).
Behavioral approaches aim to enhance self-management of dyspnea, which experts recognize as vital for improving patient well-being due to dyspnea’s persistent nature (3, 12, 13). In particular, exposure therapy—a standard treatment in anxiety disorders (14)—is used to manage symptom distress but is untested for dyspnea (15). In this case, exposure involves intentional, repeated exposures to dyspnea through physical exertion designed to provide corrective information and appropriate resilience. Exposures offer a strategy to target dyspnea-related catastrophic cognitions affecting functioning (e.g., thoughts about dying from a lack of oxygen or their inability to cope).
We developed “Take a Breath” (TAB), an exposure-based treatment for dyspnea. This effort began by formulating a cognitive-behavioral model to conceptualize dyspnea (16). It emphasized the “feedforward” cycle of symptom distress and subsequent avoidance (Figure 1). In the feedforward model, anxiety sensitivity—referring to the fear of anxiety and dyspnea-related symptoms—is central and linked to dyspnea-related distress and disability (17, 18). TAB targeted common patterns of fear and avoidance of dyspnea through psychoeducation and exposures. Patients received a finger pulse oximeter to monitor oxygen saturation and heart rate in real-time during breathlessness, which helped challenge negative thoughts (e.g., “I’m not getting enough oxygen”). TAB also encouraged physical activity through additional exposures with pulse oximetry biofeedback and activity pacing strategies (e.g., a behavioral technique encouraging patients to manage their daily activities and avoid overexertion). To address the various influences of dyspnea, TAB incorporated pursed lip breathing (19), posture changes to produce relief (20), progressive muscle relaxation technique (21), and symptom monitoring to facilitate identification of patterns of dyspnea and prompt deployment of coping strategies (22). Theory and clinical evidence supported each component; their combination is novel but untested in treating dyspnea.
A pilot randomized controlled trial examined the acceptability (treatment satisfaction), feasibility (accrual, treatment retention, and homework completion), and preliminary efficacy of TAB compared to standard of care (SOC) to improve dyspnea in patients with lung cancer. The primary outcome was dyspnea (i.e., dyspnea-related functioning, effort, anxiety, and discomfort). Secondary outcome measures were selected to map onto the cognitive-behavioral model. As an exploratory aim, we tested anxiety sensitivity as a treatment moderator. We hypothesized that TAB would be acceptable and feasible and improve dyspnea from pre- to post-treatment compared to SOC, with gains maintained at follow-up. A portion of the data was previously reported in a dissertation (1).
From May 2020 to April 2021, we recruited patients with non-small cell lung cancer (NSCLC) or small cell lung cancer (SCLC) experiencing at least moderate dyspnea (scoring ≥2 on the American Thoracic Society [ATS] Dyspnea Questionnaire), seeking care at the Ohio State University Comprehensive Cancer Center, with life expectancy >2 months, and without disabling comorbidity (e.g., dementia) affecting participation as determined by the treating oncologist.
Study personnel identified potential participants through medical chart documentation, assessed eligibility, obtained informed consent, and administered the initial assessment. An independent third party generated a random allocation sequence of two randomized blocks with a 1 allocation to TAB vs. SOC for each marital status (married vs. non-married) and current cancer treatment (any immunotherapy vs. other). The allocation sequence was concealed with sequentially numbered, password-protected computer documents. Assessments were completed online on RedCap at pre-treatment, mid-treatment (3 weeks), post-treatment (6 weeks), and at 1-month follow-up (10 weeks). Patients received a 15.
The study was approved by the Institutional Review Board following federal regulations and the Declaration of Helsinki. The trial was registered to ClinicalTrials.gov (NCT05304793) approximately one year after the first enrollment. The delay was due to a misclassification of the study as observational during our internal institutional review and was compounded by operational disruptions from the COVID-19 pandemic. The error was corrected upon identification, and we promptly registered the trial. The Appendix includes the IRB-approved study protocol used at the first enrollment and summarizes subsequent amendments. Additionally, the ClinicalTrials.gov registration refers to testing treatment tolerability, but we did not report it here as it was redundant to the preliminary efficacy analyses. Additionally, a copy-and-paste error led to the omission of the Cancer Dyspnea Scale as a measure of dyspnea in the registration.
The patient’s oncology team provided dyspnea SOC. Examples of SOC included no treatment, treatment of underlying conditions (e.g., pleural effusion), pharmacotherapy (e.g., opioids), and supplemental oxygen if hypoxemia (2, 3).
Patients randomized to TAB also received SOC. TAB was manualized and consisted of 5 one-hour weekly sessions delivered in-person or via video conferencing based on patient preference (70% of TAB sessions were delivered virtually). Sessions were completed in order and rescheduled as needed. Treatment sought to 1) target dyspnea-related thoughts (e.g., thoughts of fainting, inability to cope) through psychoeducation and interoceptive exposures paired with finger pulse oximetry biofeedback (i.e., monitoring oxygen saturation and heart rate in real-time while inducing breathlessness, helping to challenge unhelpful thoughts about their dyspnea); 2) target dyspnea through breathing retraining, postural relief, and symptom monitoring; 3) reduce aversive dyspnea-related emotions through progressive muscle relaxation; 4) promote adaptive behavior through encouraging physical activity with exposure exercises with biofeedback and activity pacing; and 5) provide a tailored plan for maintaining coping strategies post-treatment. Patients received educational handouts, worksheets, a progressive muscle relaxation exercise recording, and a consumer-grade finger pulse oximeter (Contec CMS50DL22). See Table 1 and Appendix for details on TAB.
TAB therapists were two MA-level clinical psychologists with two years of clinical experience treating patients with lung cancer. Training consisted of a review of the treatment manual, a one-hour didactic on treatment rationale and procedures, and a three-hour role-play session on treatment delivery. A licensed psychologist specializing in psychosocial oncology provided weekly supervision. Patients were assigned based on therapist availability.
TAB was considered acceptable if the treatment satisfaction mean scores were at least “mostly satisfied” (Client Satisfaction Questionnaire-8 [CSQ-8] mean ≥ 24). The CSQ-8 uses 8 items to assess patient satisfaction with treatment (23). Patients rated items on a 4-point Likert scale. Responses are summed for a score ranging from 0–32, with higher scores indicating greater treatment satisfaction.
Three variables operationalized accrual, treatment retention, and homework completion. TAB was considered feasible if 1) ≥50% of eligible patients were accrued (consented), 2) ≥70% of TAB patients (excluding deaths) completed the 5-session treatment, and 3) ≥70% of sessions having reports of completing ≥20% of the assigned homework on a homework compliance rating scale commonly used in psychotherapy studies (24). The therapist rates the scale on a 6-point Likert scale (1=did not attempt the assigned homework to 6=did more than was requested) and the percentage of homework completed for each session. See Appendix for standard homework assignments.
TAB therapists self-rated treatment fidelity by completing a checklist of major session components after each session. Provider-rated checklists are commonly used to measure fidelity in psychological interventions (25, 26). Fidelity was defined as the percentage of completed components per patient for each session.
Recognizing the multidimensional nature of dyspnea (2), four aspects of dyspnea were assessed. 1) The ATS Dyspnea Questionnaire uses 5 items to assess functional impairment due to dyspnea (27). Patients indicated if dyspnea affects functioning (e.g., stopping for breath when walking at your own pace on the level; yes=1 or no=0). Responses are summed for a total score ranging from 0–5, with higher scores indicating greater impairment. We used the reported minimally clinically important difference of 1.0 from the Medical Research Council Dyspnea Scale, a measure incorporated into the ATS Dyspnea Questionnaire (28). 2–4) The Cancer Dyspnea Scale (CDS) uses 12 items to assess dyspnea-related effort, anxiety, and discomfort (29). Patients rated breathing difficulties during “the past few days” on a 5-point Likert scale. Responses are summed for total and three subscale scores; the subscale scores were of primary interest. Ranges were 0–48 for the total score, 0–20 for effort, 0–16 for anxiety, and 0–12 for discomfort, with higher scores indicating more dyspnea-related difficulties. CDS total scores >7 correspond to clinically significant dyspnea (30).
Three measures were used. 1) The Patient Health Questionnaire-9 (PHQ-9) uses 9 items to assess the frequency of major depressive disorder symptoms over the last two weeks (31). Patients rated items on a 4-point Likert scale (0=not at all to 3=nearly everyday). Responses are summed for a score ranging from 0–27, with higher scores indicating greater depressive symptoms. We used a 5-point change as the threshold for a minimally important difference previously established for the PHQ-9 (32). 2) The Generalized Anxiety Disorder-7 (GAD-7) uses 7 items to assess the frequency of generalized anxiety disorder symptoms over the last two weeks (33). Patients rated items on a 4-point Likert scale (0=not at all to 3=nearly everyday). Responses are summed for a score ranging from 0–21, with higher scores indicating greater anxiety symptoms. We used a 5-point change as the threshold for a minimally important difference previously established for the GAD-7 (34). 3) The Profile of Mood States Short Form (POMS-SF) uses 37 items to measure negative (e.g., tense) and positive (e.g., cheerful) moods during the past week (35). Patients rated the degree to which each item describes their mood on a 5-point Likert scale (0=not at all to 4=extremely). A total mood disturbance score is calculated, ranging from −24 to 124, with a higher score indicating greater mood disturbance.
The Breathlessness Catastrophizing Scale (BCS) uses 13 items to assess the frequency of dyspnea-specific catastrophic thoughts, e.g., “I feel I can’t go on” (36). Patients rated items on a 5-point Likert scale (0=not at all to 4=all the time). Responses are summed for a total score ranging from 0–52, with higher scores indicating more dyspnea-related catastrophic thoughts. The BCS was first validated and found reliable in patients with chronic obstructive pulmonary disease and has been used in cancer (37).
The International Physical Activity Questionnaire (IPAQ) Short Form uses 7 items to assess physical activity and sedentary behavior over the past week (38). Responses are transformed to calculate the metabolic equivalent of task minute (MET-minute)—the amount of energy expended during a minute while at rest—as a measure of physical activity (39). Minutes per day reported sitting was used as a measure of sedentary behavior.
Two measures were used. 1) The Medical Outcomes Study-Short Form-12 (SF-12) uses 12 items to assess physical and mental health-related quality of life during the past month (40). Responses are calculated for a physical component summary (PCS) and mental component summary (MCS) score, each ranging from 0–100, with higher scores indicating better health-related quality of life. Scores are normed to a mean of 50 based on US population norms (41). We used a change of 6.7 and 2.4 points on the PCS and MCS, respectively, as the threshold for minimally important differences previously established for the SF-12 (42). 2) The self-rated Karnofsky Performance Status Scale (KPS) uses one item to assess functional status (43) based on the traditional provider-rated KPS (44). Patients rated their functioning on a scale ranging from 100 (“Normal, no complaints, no evidence of disease”) to 10 (“Moribund”) with 10-point intervals, each containing different criteria. Lower scores indicate more restricted performance in daily and self-care activities.
Patients self-reported their demographic information. Cancer-related information was abstracted from the medical records. Raters used medical record data to complete the Charlson Comorbidity Index (CCI) to quantify comorbidities (45). Scores range from 0–37, with higher scores indicating greater comorbidity burden and mortality risk. All received at least 2 points for having cancer.
The Anxiety Sensitivity Index (ASI-3) uses 18 items to assess anxiety sensitivity (the fear of anxiety-related sensations) related to physical, cognitive, and social concerns (46). Patients rated the degree to which anxiety symptoms are distressing on a 5-point Likert scale. Responses are summed for a score ranging from 0–72, with higher scores indicating greater anxiety sensitivity. The ASI-3 was validated and found reliable in patients with anxiety and healthy adults but has been used in respiratory disease and cancer (47, 48).
A sample of 30 was sufficient to detect a moderate-to-large effect (Cohen’s d=0.60 from a behavioral treatment for dyspnea) on dyspnea-related functioning between two groups with a power of 80% and two-tailed alpha of 5% (13). We increased the sample size to 50 to adjust for an estimated 25% treatment attrition and 10% mortality.
Missing data were examined. T-tests and Fisher’s exact tests tested for baseline differences by study arm and missingness.
The acceptability and feasibility of TAB were summarized with descriptive statistics. Robust multilevel modeling was conducted using the “robustlmm” package in R Version 4.2.1 to test the efficacy of TAB vs. SOC (49, 50). The robust estimator in multilevel modeling is based on the random effects contamination model and makes no assumptions about the grouping structure. No adjustments based on missingness were made as the estimation method utilizes patients with missing data by estimating the best-fitting model from the observed data (51, 52). Intention-to-treat and per-protocol analyses were completed to estimate a conservative and upper-bound treatment effect. For each outcome, multilevel modeling estimated effects for Time (fixed; pre-treatment, mid-treatment, post-treatment, and 1-month follow-up), Group (fixed; TAB=1 and SOC=0), the Group × Time interaction, and the random intercept by subjects. Treatment effect sizes were calculated by dividing the difference of the estimated marginal means of the 2 groups by the baseline standard deviation (53). Sensitivity analyses found models unaffected by the inclusion of covariates (patient characteristics correlated with outcomes) and thus were omitted. Models included continuous time (i.e., days from enrollment to time of assessment) as an additional fixed effect when the inclusion provided a better-fitting model for an outcome. To better contextualize our primary outcome, we used a Chi-Square test to compare the proportion of patients in each group reporting at least a minimally important difference on the ATS Dyspnea Scale by post-treatment. We then used a post-hoc logistic regression to calculate an odds ratio, controlling for the baseline outcome.
Moderation effects of baseline anxiety sensitivity were evaluated by including the interaction of the mean-centered moderator with the Group × Time interaction for the ITT multilevel models of the primary outcomes. The “lme4” and “interactions” R packages probed the interaction with Johnson-Neyman intervals (54, 55).
See Figure 2 for the CONSORT diagram of the study flow. Fifty patients consented to participate, of which four either did not meet inclusion criteria or withdrew before study registration, resulting in 46 stratified and randomized to TAB (n=22) or SOC (n=24). One patient in the SOC group withdrew before data collection, leaving a total analytic sample of 45 patients assigned to TAB (n=22) or SOC (n=23).
The sample was predominantly of older age (Mean=64.0±8.2), non-Hispanic White (88.9%), unmarried (51.1%), and had more than a high school education (71.1%). Most were in treatment (91.1%) for their stage IV NSCLC (80.0%) or extensive stage SCLC (13.3%). As anticipated, patients reported clinically significant dyspnea on the multidimensional dyspnea measure (n=38; 84.4%; CDS Total M=15.8±7.6). At baseline, patients, on average, reported dyspnea to cause them to stop for breath when walking at their own pace on the level (ATS Dyspnea M=3.8±1.1). There were high reported levels of dyspnea-related anxiety (CDS anxiety M=4.3±3.2), discomfort (CDS discomfort M=3.8±2.1), and effortful breathing (CDS effort M=7.6±3.7). Importantly, the study arms did not differ in any baseline patient characteristic or study measure (ps>0.086). See Table 2.
Thirteen patients had missing data (28.9%). Missingness did not differ by study arm (8 in TAB vs. 5 in SOC; p=0.34). Patients with missing data were more likely to have SCLC vs. NSCLC (30.7% vs. 6.3%; p=0.05) and be sedentary (M=709 vs. 521 min/day; p=0.02).
TAB had high treatment satisfaction scores on the CSQ (M=27.4 ± 6.28; range=11–32). Of those who reported treatment satisfaction, most reported being at least “mostly satisfied” with TAB (n = 12; 75%). Three metrics assessed “feasibility.” Of the 195 potentially eligible patients, 50 (25.6%) were accrued. Regarding homework completion, TAB homework adherence was high, with 88.7% of patients completing ≥20% of the assignments each session. Regarding treatment retention, 70% (n=14) of the patients completed the primary treatment components (≥3 sessions), whereas 60% completed the full treatment, excluding the two patients referred to hospice during the study. See the Appendix for details on retention rates, treatment fidelity ratings, and homework adherence. Fidelity ratings were >90% for all sessions.
In preface, all results described below are Group × Time interactions. Additionally, all outcomes noted were maintained from post-treatment to follow-up based on non-significant interactions. See Table 3 and Figure 3 for the summary of results.
The TAB arm had large improvements in dyspnea outcomes relative to SOC. Dyspnea-related anxiety (CDS anxiety) improved from pre-treatment to post-treatment (b=−2.17, p<0.01, Cohen’s d=0.87) and to follow-up (b=−1.65, p<0.01, Cohen’s d=0.71). Additionally, TAB improved dyspnea-related functional impairment (ATS dyspnea) from pre- to post-treatment (b=−0.74, p=0.03, Cohen’s d=0.82). The percentage of patients who improved on dyspnea-related functional impairment from baseline to post-treatment by at least 1.0 point (the minimally important difference) was higher in TAB (50.0% vs. 26.1%; Chi-Square[df] = 3.91[1], p = 0.05). A post-hoc logistic regression showed that patients receiving TAB were more likely to have a clinically significant difference in dyspnea-related functioning compared to the control after controlling for baseline scores (OR=4.85, 95% CI = 1.20 – 22.42, p=0.03)
TAB had greater improvement than SOC with mental health-related quality of life (SF-12 MCS) from pre-treatment to post-treatment (b=10.45, p<0.01, Cohen’s d=0.57) and continued to improve into follow-up (b=11.19, p<0.01, Cohen’s d=0.64). Correspondingly, TAB improved depressive symptoms (PHQ-9) from pre-treatment to post-treatment (b=−2.70, p=0.04, Cohen’s d=0.41) and follow-up (b=−3.35, p=0.01, Cohen’s d=0.54) compared to SOC.
Compared to SOC, TAB improved functional status (KPS) from pre-treatment to follow-up (b=8.39, p=0.02, Cohen’s d=1.08). Time spent sedentary also improved pre- to post-treatment (b=−145.30, p=0.02, Cohen’s d=0.23) more than SOC; patients receiving TAB spent 2.5hrs/day less sedentary than those receiving SOC by post-treatment. There was no significant Group × Time interaction by post-treatment for physical health-related quality of life (SF-12 PCS; p=0.44, Cohen’s d=0.10). However, patients receiving TAB continued to improve from post-treatment to follow-up at a greater rate than SOC (b=5.57, p=0.02, Cohen’s d=0.85).
Per-protocol analyses demonstrated findings similar to the ITT analyses favoring TAB over SOC (see Table 3).
Anxiety sensitivity moderated the effect of TAB on dyspnea-related functioning (b=−0.001, 95% CI = −0.002 - −0.001, p<0.01, Cohen’s d=0.31). Patients with higher baseline anxiety sensitivity benefited more from TAB over time. The improvement rate significantly increased when baseline anxiety sensitivity exceeded 7 (indicating ≥ “almost no anxiety sensitivity” per ASI-3 cutoffs), which included 64% of the sample. Anxiety sensitivity did not moderate the effect on CDS. See the Appendix for a figure depicting the moderation.
TAB was acceptable, with 75% of patients reporting to be at least “mostly satisfied.” TAB was feasible based on homework adherence rates (88.7%) and retention rates of completing the primary treatment components after excluding patients referred to hospice while on study (70%). TAB was not feasible based on accrual (25.6%). TAB yielded large improvements in dyspnea-related functioning (d=0.82) and anxiety (d=0.87) compared to the control in the ITT analyses. Relative to the control, TAB also improved secondary outcomes, including performance status, sedentary time, and depressive symptoms (d’s = 0.23–1.08).
This study provides initial evidence for the acceptability, feasibility, and efficacy of a 5-session cognitive behavioral intervention for dyspnea. It is the first to combine exposure-based interventions with oximetry biofeedback for dyspnea designed to enhance self-efficacy in its management (56). Notably, the magnitudes of TAB’s effects on dyspnea were four times larger than those reported in recent meta-analyses of opioids (d=0.14) and exercise interventions (d=0.18) (6, 57). TAB roughly doubled the likelihood for clinically significant improvements to dyspnea-related functioning compared to the control (50.0% vs. 26.1%), indicating promising preliminary efficacy for a low-risk supportive care intervention to improve functioning affected by a persistent symptom like dyspnea. TAB also improved dyspnea-related anxiety without significant changes to general anxiety (e.g., GAD-7) or sensory-perceptual dimensions of dyspnea, suggesting a targeted effect. We can only speculate on the clinical importance of TAB’s impact on dyspnea-related anxiety without data on the measure’s minimally important differences. However, the large observed effect, coupled with improvements in key secondary outcomes, leads us to hypothesize addressing dyspnea-related anxiety and fostering a sense of mastery over dyspnea may enhance patients’ mental and physical well-being.
Improvements in dyspnea through skill enhancement may help patients return to activities and reduce physical limitations, as patients receiving TAB spent 2.5 fewer hours/day sedentary post-treatment than the control group. This increase aligns with improved self-reported performance status at follow-up, which is noteworthy, as returning to activity is a top priority for patients (58, 59). TAB had a five-fold improvement on the minimally important difference in mental health-related quality of life post-treatment and an improvement just below the minimally important difference in physical health from post-treatment to follow-up (42). Driving the change in mental health-related quality of life could be the moderate effect of TAB on depressive symptoms, which corresponded to half the minimally important difference in the PHQ-9 (32). As patients live longer with new therapies, TAB’s impact on physical and mental health outcomes is more meaningful.
We identified anxiety sensitivity as an important individual difference that moderates our exposure-based treatment for dyspnea. Corresponding to its association with worse respiratory outcomes and healthcare utilization in patients with lung disease (60), our results support our hypothesis that patients with high anxiety sensitivity may experience worse dyspnea-related disability that may be due to an overreliance on avoidance behaviors (inactivity) to manage dyspnea (61). Understanding and targeting anxiety sensitivity is crucial due to its potential influence on dyspnea-related disability.
The high treatment satisfaction reports are noteworthy, considering that exposures could be aversive to patients. Further, the accruing 50 patients within a year reflected a high demand for dyspnea care. Our accrual relied on screening for dyspnea in the medical records, overestimating the number of potentially eligible patients and, consequently, underestimating the accrual rate. Regarding retention, TAB was similar to other psychosocial interventions in advanced cancer, which range from 35–100% (62). The retention of 60% was notable, considering barriers to care in lung cancer (e.g., multiple medical appointments) and patients’ poor health. Non-completers had worse pre-treatment dyspnea-related functioning, half of which cited a desire to focus on cancer treatment or other health issues as their reason for non-completion. It is also unknown how the height of the COVID-19 pandemic affected treatment completion. Future research could focus on treatment adaptations to improve retention.
The limitations and context of the study are considered. The sample detected treatment effects and we used sensitivity analyses to determine if results differed by analytic approaches. However, the small sample could overestimate effect sizes and a larger trial is needed to determine efficacy. Additionally, the retention rate of 60% is a limitation, and efforts to test methods to improve retention would improve the potential clinical impact of TAB. The study continued accrual and treatment delivery during the COVID-19 pandemic with virtual delivery as needed, providing an additional perspective on the effects observed. The cognitive-behavioral model guided the identification of theoretically relevant measures, which, although relying on self-report, better reflected patients’ experiences. Therapists were assigned based on availability, which may have confounded results. Therapists also self-reported fidelity, which may have biased the ratings. We found the sample to be statistically balanced at baseline, but the control group showed signs of worse disease (e.g., non-significantly longer time since cancer diagnosis). We observed no group differences in our baseline outcomes, including dyspnea, though sample imbalance could confound results. The sample was comparable in sex, disease stage, and age to nationally representative populations with lung cancer, but not for race/ethnicity (63, 64). The sample was recruited from a single site, disproportionately White, diagnosed with lung cancer, and was well enough to receive outpatient cancer care, limiting generalizability to other settings, samples, and disease groups.
In conclusion, we provide evidence for the acceptability and feasibility of a novel psychological intervention for dyspnea outcomes. TAB is the first comprehensive cognitive-behavioral treatment with preliminary efficacy in improving dyspnea and related sequelae. TAB is theoretically grounded, evidence-based, and manualized—providing a firm basis for adaptation to other disease groups and treatment settings. These early data support the need for a follow-up, larger trial to establish efficacy, which is much needed given the high demand for supportive cancer care (65).