Background
In addition to the health threat posed by the acute coronavirus disease 2019 (COVID-19), long-term consequences present as another major health concern of the COVID-19 pandemic. Whereas most cases fully recover from acute illness, a considerable proportion reports the persistence of multiple physical and mental complaints. The post-COVID-19 condition (PCC) is defined as the experience of symptoms at least 12 weeks after a SARS-CoV-2 infection that developed during acute COVID-19 or after initial recovery, persist for at least two months, interferes with everyday functioning, and cannot be explained by an alternative diagnosis (Soriano, Murthy, Marshall, Relan, & Diaz, Reference Soriano, Murthy, Marshall, Relan and Diaz2022). Common symptoms include fatigue, headache, cognitive impairments, or dyspnea, but overall, PCC can affect multiple organ systems resulting in a variety of associated symptoms and different phenotypes (Groff et al., Reference Groff, Sun, Ssentongo, Ba, Parsons, Poudel and Chinchilli2021; Lopez-Leon et al., Reference Lopez-Leon, Wegman-Ostrosky, Perelman, Sepulveda, Rebolledo, Cuapio and Villapol2021; Reese et al., Reference Reese, Blau, Casiraghi, Bergquist, Loomba, Callahan and Robinson2023). A meta-analysis estimated the global pooled prevalence to be 43% of people with confirmed COVID-19 diagnosis, but prevalence estimates vary widely by study procedure (Chen et al., Reference Chen, Haupert, Zimmermann, Shi, Fritsche and Mukherjee2022). More conservative estimates suggest that approximately 1–5% develop somatic complaints in accordance with PCC criteria (Thompson et al., Reference Thompson, Williams, Walker, Mitchell, Niedzwiedz, Yang and Steves2022). Either way, given the number of people infected with SARS-CoV-2 worldwide, PCC affects a significant number of people, resulting in increased health care burden and socioeconomic costs (Tartof et al., Reference Tartof, Malden, Liu, Sy, Lewin, Williams and Qian2022; Williamson, Tydeman, Miners, Pyper, & Martineau, Reference Williamson, Tydeman, Miners, Pyper and Martineau2022).
So far, the pathogenesis of PCC is unknown. It is most probably multifactorial and differs between patients. Several biomedical explanation attempts are currently discussed (Merad, Blish, Sallusto, & Iwasaki, Reference Merad, Blish, Sallusto and Iwasaki2022), although a considerable proportion of patients do not exhibit biomedical abnormalities (Fleischer et al., Reference Fleischer, Szepanowski, Tovar, Herchert, Dinse, Schweda and Kleinschnitz2022; Reese et al., Reference Reese, Blau, Casiraghi, Bergquist, Loomba, Callahan and Robinson2023; Sneller et al., Reference Sneller, Liang, Marques, Chung, Shanbhag, Fontana and Lane2022). Persistent somatic symptoms (PSS) that cannot be sufficiently explained by biomedical factors are common across other health conditions and often better represented by biopsychosocial models (Engel, Reference Engel1977; Van den Bergh, Witthöft, Petersen, & Brown, Reference Van den Bergh, Witthöft, Petersen and Brown2017). Interactions between biological and psychosocial factors also seem relevant in PCC (Lemogne, Gouraud, Pitron, & Ranque, Reference Lemogne, Gouraud, Pitron and Ranque2023; Saunders, Sperling, & Bendstrup, Reference Saunders, Sperling and Bendstrup2023). For example, elevated psychological distress was seen to be both a symptom and a risk factor of PCC (Dong, Liu, Dai, Yang, & Liu, Reference Dong, Liu, Dai, Yang and Liu2021; Magnúsdóttir et al., Reference Magnúsdóttir, Lovik, Unnarsdóttir, McCartney, Ask, Kõiv and Pedersen2022). Further, observational studies have shown that individuals with PCC are at risk for DSM-5 Somatic Symptom Disorder (SSD) indicating the presence of psychobehavioral features (e.g. catastrophizing thoughts, health anxiety, specific illness behaviors) that might interfere with successful coping of PCC and thus contribute to somatic symptom distress (Horn et al., Reference Horn, Wathelet, Amad, Martignène, Lathiere, Khelfaoui and D'Hondt2023; Kachaner et al., Reference Kachaner, Lemogne, Dave, Ranque, de Broucker and Meppiel2022; Schneider et al., Reference Schneider, Huber, Lohse, Linde, Greissel, Sattel and Hapfelmeier2023; Willis & Chalder, Reference Willis and Chalder2021).
Based on treatment suggestions for PSS (e.g. Henningsen, Zipfel, & Herzog, Reference Henningsen, Zipfel and Herzog2007), guidelines for PCC mostly recommend multidisciplinary rehabilitation strategies focusing on self-management (Koczulla et al., Reference Koczulla, Ankermann, Behrends, Berlit, Berner, Böing and Zwick2022; Shah, Hillman, Playford, & Hishmeh, Reference Shah, Hillman, Playford and Hishmeh2021). In Germany, existing rehabilitation programs have been adapted to the treatment of PCC and are currently evaluated (Kupferschmitt et al., Reference Kupferschmitt, Hinterberger, Montanari, Gasche, Hermann, Jöbges and Köllner2022). So far, these programs have been shown to be beneficial for PCC patients, and effects may be comparable to those of other rehabilitation cohorts (e.g. psychocardiology patients; Kupferschmitt et al., Reference Kupferschmitt, Langheim, Tüter, Etzrodt, Loew and Köllner2023). A common component of multidisciplinary rehabilitation programs, particularly those aimed at self-management of PSS, are psychological interventions such as cognitive-behavioral therapy (CBT). CBT protocols for PSS typically include interventions that address common psychophysiological interactions and promote adaptive coping strategies (e.g. stress management, cognitive restructuring, behavioral activation; Witthöft & Hiller, Reference Witthöft and Hiller2010), yielding small to medium-sized effects on the reduction of somatic symptom severity including fatigue (Ingman, Smakowski, Goldsmith, & Chalder, Reference Ingman, Smakowski, Goldsmith and Chalder2022; Kleinstäuber, Witthöft, & Hiller, Reference Kleinstäuber, Witthöft and Hiller2011; Price, Mitchell, Tidy, & Hunot, Reference Price, Mitchell, Tidy and Hunot2008; Van Dessel et al., Reference Van Dessel, den Boeft, van der Wouden, Kleinstäuber, Leone, Terluin and van Marwijk2014). Despite promising first results (Frisk et al., Reference Frisk, Jürgensen, Espehaug, Njøten, Søfteland, Aarli and Kvale2023; Kuut et al., Reference Kuut, Müller, Csorba, Braamse, Aldenkamp, Appelman and Knoop2023), the application of CBT principles for the treatment of PCC is still under research.
The present study aims to examine the feasibility of a specific CBT protocol for the treatment of PCC. Therefore, we developed a comprehensive CBT-based group therapy program (CBT-PCC) that addresses common challenges of coping with PSS. The feasibility, acceptance, and effectiveness were evaluated in a cohort of inpatients fulfilling PCC criteria in a neurological rehabilitation setting.
Methods
Participants and procedure
Participants were recruited during inpatient neurological rehabilitation for PCC between February and July 2022 in the Kliniken Schmieder Konstanz and Gailingen, Germany. On admission, patients were screened for eligibility (including PCC assessment) and informed about CBT-PCC by a clinic physician during a medical interview. Inclusion criteria were the fulfillment of PCC criteria according to the WHO (i.e. history of probable or confirmed SARS-CoV-2 infection, at least one symptom after ⩾3 months post-infection, symptom persistence for ⩾2 months, symptom(s) cannot be explained by an alternative diagnosis, impact on everyday functioning; Soriano et al., Reference Soriano, Murthy, Marshall, Relan and Diaz2022), age of at least 18 years, literacy in German, and written informed consent. History of SARS-CoV-2 infection was confirmed based on medical documentation by the referring physician. Exclusion criteria were mechanical ventilation or admission to an intensive care unit during acute COVID-19 and a diagnosis of schizophrenia spectrum disorders or acute substance addiction.
This study was conducted as a single-group, uncontrolled clinical trial. After each session, participants and therapists completed session evaluation questionnaires. Psychometric assessment of somatic and psychopathological symptom burden occurred pre- and post-intervention (i.e. after a maximum of four weeks or at discharge from inpatient rehabilitation). Additionally, a structured clinical diagnostic interview to examine mental disorders was added to pre-intervention assessment. Post-intervention assessment contained a standardized measure to examine adverse events.
The trial was retrospectively registered at the German Clinical Trials Register (www.drks.de, DRKS00031219). The study protocol was reviewed and approved by the ethics board of the University of Konstanz (approval number 44/2021, December 13, 2021). All participants provided written informed consent prior to the initiation of any study procedures. Refusal to participate or discontinuation of the study did not result in any negative consequences for the inpatient rehabilitation treatment.
Treatment
The group intervention comprised eight modules (à 60 min, over four weeks) and was offered additionally to routine inpatient neurological rehabilitation (e.g. medical consultation, physiotherapy, medical training therapy, occupational therapy, cognitive training, individual psychotherapeutic sessions). The treatment protocol was based on an established CBT manual for SSD (Kleinstäuber, Thomas, Witthöft, & Hiller, Reference Kleinstäuber, Thomas, Witthöft and Hiller2018), which has been shown to effectively reduce somatic symptom distress in recent trials (Hennemann et al., Reference Hennemann, Böhme, Kleinstäuber, Baumeister, Küchler, Ebert and Witthöft2022; Kleinstäuber et al., Reference Kleinstäuber, Allwang, Bailer, Berking, Brünahl, Erkic and Rief2019). The suggested interventions of the treatment protocol were reviewed in advance by practitioners from PCC rehabilitation and transcribed in a study manual that was provided to the study therapists. Modules were included in the following order: (1) introduction and goal setting, (2) psychoeducation, (3) stress education and relaxation, (4) attention modification, (5) cognitive restructuring, (6) balancing physical activity, (7) stress management, and (8) summary and transfer. Each module began with a breathing exercise and ended with the assignment of homework. Because the original treatment manual was not tailored to a specific somatic symptom cluster, we largely maintained the transsymptomatic character to ensure appropriateness for a wide range of PCC phenotypes and other treatment settings during the development of CBT-PCC. Nonetheless, PCC-specific modifications were implemented to some extent, most notably through a particular emphasis on breathing exercises, PCC-specific psychoeducation (e.g. appearance, course, epidemiology), and session modalities (i.e. duration, amount of content). Content and therapeutic strategies of CBT-PCC are shown in Table 1.
Table 1. Content and therapeutic strategies of the group program
Note. PCC, Post-COVID-19 condition.
Three licensed psychotherapists and two psychologists in neuropsychological training that were trained in a 1-day workshop (provided by DH and MW) delivered the intervention. In addition, a training cycle of the group program was conducted prior to data collection.
Measures
Session evaluations
For session evaluations, we used the Group Therapy Session Evaluation form (GTS; Rief, Bleichhardt, & Timmer, Reference Rief, Bleichhardt and Timmer2002; Zoubek, Reference Zoubek2013). The GTS is a German self-report instrument to assess evaluations of therapeutic processes such as satisfaction, personal involvement, perceived usefulness, atmosphere, and comprehensibility in patients (GTS-P) and therapists (GTS-T). The patient version comprises eight items that are rated on a 5-point scale (0 = ‘disagree’ to 4 = ‘agree’). In the study sample, the GTS-P showed high internal consistencies across sessions (α = 0.85–0.94). For the therapist version, items were reformulated in terms of how therapists rate patients’ perceptions of the same aspects. In addition, items are included to assess therapists’ satisfaction with the session, as well as ratings of how well the protocol is applicable and meets patients’ needs. The GTS-T comprises 12 items, each rated on the same 5-point scale and showed high internal consistencies (α = 0.85–0.96) as well. Higher scores reflect a more positive judgment of the session, except for item 9 of the GTS-T, which is inverted.
Adverse events
Adverse events during the intervention were monitored using the Unwanted Events in Group Therapy Scale (UE-G; Linden, Walter, Fritz, & Muschalla, Reference Linden, Walter, Fritz and Muschalla2015). It includes 47 events regarding different domains (i.e. room situation, session content, participant behavior, therapist behavior, undesirable long-term effects, global evaluation) that are rated on a 5-point scale (0 = ‘did not occur’ to 4 = ‘did occur and was extremely burdensome’).
Somatic symptom distress and fatigue
Somatic symptom distress was measured using the somatic symptom scale of the Patient Health Questionnaire (PHQ-15; Kroenke, Spitzer, & Williams, Reference Kroenke, Spitzer and Williams2002) and the Somatic Symptom Disorder – B criteria scale (SSD-12; Toussaint et al., Reference Toussaint, Murray, Voigt, Herzog, Gierk, Kroenke and Löwe2016). The PHQ-15 is a validated self-report screening that assesses the severity of 15 common somatic symptoms during the last four weeks on a 3-point scale (0 = ‘not bothered at all’ to 2 = ‘bothered a lot’). The total score ranges between 0 and 30 points and scores ⩾5, ⩾10, and ⩾15 can be considered cut-off points for mild, medium, and severe somatic symptom severity, respectively. The SSD-12 measures psychobehavioral features of somatic symptom distress according to the B criterion of SSD. The 12 items are rated on a 5-point scale (0 = ‘never’ to 4 = ‘very often’) and can be aggregated to a global sum score (range 0 − 48).
Because fatigue represents a core symptom of PCC, subjective fatigue was measured more specifically using the Fatigue Severity Scale (FSS; Krupp, LaRocca, Muir-Nash, & Steinberg, Reference Krupp, LaRocca, Muir-Nash and Steinberg1989). The FSS is a self-report questionnaire and comprises nine statements that are rated on a 7-point scale (1 = ‘does not apply’ to 7 = ‘does fully apply’) and averaged for a composite score. The proposed cut-off for clinically relevant fatigue severity is ⩾4. The internal consistencies in this study were Cronbach's α = 0.77, 0.89, and 0.94 for PHQ-15, SSD-12, and FSS, respectively.
Depression and anxiety
Depressive symptoms were measured using the depression scale of the Patient Health Questionnaire (PHQ-9; Kroenke and Spitzer, Reference Kroenke and Spitzer2002) that includes nine symptoms of depression that are rated for the last two weeks on a 4-point scale (0 = ‘not at all’ to 3 = ‘nearly every day’, range 0 − 27). The proposed cut-off for screening for depression is ⩾10. Anxiety symptoms were measured with the Generalized Anxiety Disorder Questionnaire (GAD-7; Spitzer, Kroenke, Williams, & Löwe, Reference Spitzer, Kroenke, Williams and Löwe2006) that includes seven symptoms of general anxiety that are rated for the last two weeks on a 4-point scale (0 = ‘not at all’ to 3 = ‘nearly every day’, range 0 − 21). The proposed cut-off for screening for anxiety disorders is ⩾10. Internal consistencies in the present study were Cronbach's α = 0.80 and 0.92 for PHQ-9 and GAD-7, respectively.
Illness-related cognitions
For the assessment of illness-related cognitions, the Illness Cognition Questionnaire (ICQ; Evers et al., Reference Evers, Kraaimaat, van Lankveld, Jongen, Jacobs and Bijlsma2001) was administered. The ICQ comprises 18 statements regarding cognitions of helplessness, acceptance, and perceived benefits of chronic diseases that are rated on a 4-point scale (1 = ‘not at all’ to 4 = ‘completely’; subscale range 6 − 24). Further, illness-related self-efficacy was measured using the self-efficacy module of the Hamburg Modules for the Assessment of Psychosocial Health in Clinical Practice (HEALTH-49; Rabung et al., Reference Rabung, Harfst, Kawski, Koch, Wittchen and Schulz2009). It comprises five items that are rated for the last two weeks on a 5-point scale (0 = ‘not at all’ to 4 = ‘very much’) and are averaged for a composite score. In the current study, the internal consistencies of the ICQ subscales varied between Cronbach's α = 0.80 and 0.91. For the self-efficacy module of the HEALTH-49, internal consistency was Cronbach's α = 0.85.
Additional measures
Additional measures included a sociodemographic questionnaire assessing age, gender, education, and employment status at baseline. COVID-19-related information such as infection date was retrieved from baseline medical documentation. Because symptom onset was not reported, the difference between infection date and study enrollment served as a proxy for PCC duration. Diagnoses of mental disorders were assessed using the German version of the Structured Clinical Interview for DSM-5 Disorders (SCID-5; Beesdo-Baum, Zaudig, & Wittchen, Reference Beesdo-Baum, Zaudig and Wittchen2019).
Statistical analyses
We used IBM SPSS version 23 (SPSS Inc., Chicago, Illinois) for data analysis. For session evaluations, we calculated item means and standard deviations of both GTS versions for every session. For pre-post comparisons of somatic and psychopathological symptom burden, we conducted paired samples t tests and calculated within-group effects as d av (Lakens, Reference Lakens2013) with the intention-to-treat (ITT) sample and a per-protocol subsample of participants who completed pre- and post-intervention assessments and were adherent to the treatment (i.e. attended ⩾6 sessions). Missing data at post-intervention assessment were handled using the last observation carried forward (LOCF) method. For adverse events, frequencies and item means of the UE-G were calculated. All tests were two-sided with a significance level of 0.05.
Results
Participant characteristics
Figure 1 shows the flow of participants through the trial. Between February 1, 2022, and July 7, 2022, 90 patients admitted to neurological inpatient rehabilitation for PCC were screened for eligibility, of which 64 patients were positively screened and gave informed consent to participate in the study. One participant did not complete pre-intervention assessment and 22 participants did not complete post-intervention assessment. Participants who attended at least one session (n = 51 [79.7%]) completed on average 6.43 (s.d. = 1.80) out of eight modules. Thirteen (20.3%) participants did not attend any session of CBT-PCC. The most common dropout reasons were scheduling conflicts with other treatment components and early discharge from inpatient rehabilitation. Regarding baseline sociodemographic and clinical characteristics, participants who adhered to the study protocol (n = 32 [50.0%]; i.e. completion of pre- and post-intervention assessments and attendance of ⩾6 sessions) reported significantly less somatic (PHQ-15; M (s.d.)Per-protocol = 11.88 (4.69), M (s.d.)Drop-out = 14.74 (4.96), t(61) = 2.36, p = 0.022), depressive (PHQ-9; M (s.d.)Per-protocol = 9.31 (4.65), M (s.d.)Drop-out = 13.32 (4.96), t(61) = 3.31, p = 0.002), and anxiety symptoms (GAD-7; M (s.d.)Per-protocol = 4.84 (4.77), M (s.d.)Drop-out = 7.48 (5.30), t(61) = 2.08, p = 0.042), as well as less helplessness (ICQ; M (s.d.)Per-protocol = 14.13 (3.26), M (s.d.)Drop-out = 16.50 (4.46), Welch's t(53.05) = 2.36, p = 0.022) than participants who discontinued study and/or intervention participation. No further significant differences were observed between per-protocol participants and dropouts in terms of baseline characteristics (ps > 0.05; see online Supplementary Table A.1).
Figure 1. Participant flow. ITT, intention-to-treat. Participants were included in the per-protocol analysis if they completed pre-post assessments and attended ⩾6 sessions.
Participants (n = 63) were predominantly female (51 [81.0%]) with a mean age of 47.05 years (s.d. = 11.45, range 22 − 63). At baseline, symptom prevalence measured with the PHQ-15 (ratings ⩾1) was highest for feeling tired or having low energy (96.8%), headaches (88.8%), trouble sleeping (84.1%), pain in arms, legs, or joints (80.7%), and shortness of breath (79.4%). Forty-seven (74.6%) participants experienced medium-to-high levels of somatic symptom severity, and 31 (49.2%) participants were at risk for SSD using a combined cutoff score of PHQ-15 (⩾9) and SSD-12 (⩾23; Toussaint, Hüsing, Kohlmann, and Löwe, Reference Toussaint, Hüsing, Kohlmann and Löwe2020). Moreover, 57 (90.5%), 37 (58.7%), and 14 (22.2%) showed clinically relevant levels of fatigue, depressive, and anxiety symptoms, respectively. The mean time between first SARS-CoV-2 infection and study participation was 62.45 weeks (s.d. = 27.67, range 13.29 − 121.14). Table 2 summarizes sociodemographic and clinical characteristics of the sample.
Table 2. Sociodemographic and clinical characteristics of the sample at baseline
Note. an = 62, bn = 61, cn = 60. PHQ-15, Patient Health Questionnaire – somatic symptom scale; SSD-12, Somatic Symptom Disorder – B criteria scale; FSS, Fatigue Severity Scale; PHQ-9, Patient Health Questionnaire – depression scale; GAD-7, General Anxiety Disorder Scale; HEALTH-49, Hamburg Modules for the Assessment of Psychosocial Health; ICQ, Illness Cognition Questionnaire.
The SCID-5 was conducted with 61 participants. Thirty-three (54.1%) participants were diagnosed with at least one current mental disorder. The most frequent disorder categories were depressive disorders (n = 16), somatic symptom and related disorders (n = 15) and anxiety disorders (n = 11). A single mental disorder was diagnosed in 20 (32.8%) participants. A second comorbidity was present in ten (16.4%), and a third or fourth comorbidity was present in three (4.9%) participants. When past mental disorders are also considered, the proportion of participants with at least one mental disorder during lifetime increases to 60.7% (n = 37). Frequencies of all assigned SCID-5 diagnoses are depicted in online Supplementary Table A.2.
Session evaluations
Session evaluations by the participants (GTS-P) are listed in Table 3. The mean ratings per session were high across all sessions (all total means ⩾3, possible range 0 − 4) and varied from 3.01 (s.d. = 0.95, module 8) to 3.49 (s.d. = 0.58, module 6). Satisfaction with the session was high for every session (item 8: all means ⩾3). Participants were actively engaged (items 1 − 2: all means ⩾3) and rated every session as comprehensible (item 3: all means ⩾3). The perceived usefulness for disease coping, especially for modules 1, 2, and 8, was rated on average slightly lower (see items 4 − 6).
Table 3. Session evaluation by participants
Note. an = 47, bn = 45, cn = 44, dn = 41, en = 33. GTS-P, group therapy session evaluation by patients. Items were rated on a 5-point scale from (0) „disagree” to (4) „agree”.
The average session evaluation ratings by the therapists (GTS-T) varied more between sessions compared to participants, ranging from 2.08 (s.d. = 0.56) in module 3 to 3.54 (s.d. = 0.48) in module 8. Therapists’ ratings of their satisfaction were comparatively lower for sessions 1 to 3 (item 12: each mean <3), but eventually increased to an equivalent level for sessions 4 to 8 (each mean ⩾3). Therapists’ ratings of comprehensibility (item 3) and applicability (item 10) of each module followed a similar trend. Detailed GTS-T ratings are shown in online Supplementary Table A.3.
Adverse events
At least one adverse event was reported by 85.4% of the participants (i.e. at least one UE-G item ⩾1), 29.3% reported severe and extremely distressing events (i.e. at least one UE-G item ⩾3). The highest item mean scores were observed for ‘I realized how complicated everything is’ (M = 1.10, s.d. = 1.24), ‘I was afraid not to know how to proceed in the future’ (M = 0.78, s.d. = 1.11), and ‘I felt that my problems are more severe than I thought before’ (M = 0.71, s.d. = 1.17). Results for all events are shown in online Supplementary Table A.4.
Pre-post comparisons
Results of the ITT analysis (n = 63) are depicted in Table 4. Pre-post differences showed a significant reduction in subjective fatigue (FSS: t(62) = −2.60, p = 0.012, d av = 0.33), and helplessness (ICQ: t(60) = −2.99, p = 0.004, d av = 0.39), as well as increased acceptance (ICQ: t(60) = 3.31, p = 0.002, d av = 0.43), perceived benefits ratings (ICQ: t(60) = 3.78, p < 0.001, d av = 0.49), and illness-related self-efficacy (HEALTH-49: t(61) = 2.61, p = 0.011, d av = 0.33). Per-protocol analysis (n = 32; see online Supplementary Table A.5) showed similar results with descriptively slightly larger effect sizes for these measures (FSS: t(31) = −2.70, p = 0.011, d av = 0.48; ICQHelplessness: t(30) = −3.53, p = 0.001, d av = 0.64; ICQAcceptance: t(30) = 3.97, p < 0.001, d av = 0.72; ICQBenefits: t(30) = 4.13, p < 0.001, d av = 0.75; HEALTH-49: t(30) = 2.41, p = 0.022, d av = 0.43). The other measures did not yield any significant changes in any of the analyses.
Table 4. Pre-post comparisons of self-report measures for the intention-to-treat sample
Note. an = 62, bn = 61. PHQ-15, Patient Health Questionnaire – somatic symptom scale; SSD-12, Somatic Symptom Disorder – B criteria scale; FSS, Fatigue Severity Scale; PHQ-9, Patient Health Questionnaire – depression scale; GAD-7, General Anxiety Disorder Scale; HEALTH-49, Hamburg Modules for the Assessment of Psychosocial Health; ICQ, Illness Cognition Questionnaire.
Discussion
Identifying effective treatments is important to reduce somatic symptom burden and restore premorbid levels of functioning of people who developed long-term health sequelae of COVID-19. So far, non-pharmacological interventions including psychological interventions are among the most promising treatment options and are currently implemented by health care providers based on existing multidisciplinary rehabilitation approaches (Kupferschmitt et al., Reference Kupferschmitt, Hinterberger, Montanari, Gasche, Hermann, Jöbges and Köllner2022). However, the evidence on the efficacy, acceptability, and safety of specific treatment options is scarce. The present study examined the feasibility of a comprehensive CBT-based group therapy program targeting psychophysiological mechanisms of somatic symptom distress in a sample of inpatients with PCC.
Most importantly, our results support that CBT-PCC is feasible, well accepted, and potentially effective as part of inpatient neurological rehabilitation. In particular, participants’ evaluations of the therapeutic process were on average very positive across all sessions and the module completion rate among those who received the intervention was substantial. Participants’ ratings of the comprehensibility and atmosphere of the sessions were consistently positive, and their active engagement was high. Similarly, the assessment of the sessions’ usefulness for disease coping was good. The fact that ratings were comparatively lower for sessions 1, 2, and 8 is in line with the structure of the treatment protocol, which promotes adaptive coping strategies particularly in sessions 3 to 7. The therapists' evaluations of the treatment protocol varied more between sessions, starting at a moderate level until eventually matching the participants' evaluations from mid-intervention onward. For adverse effects, we chose a rather sensitive measure resulting in higher rates of unwanted events at first sight. However, rates are comparable to other trials evaluating CBT-based group therapy programs with the same instrument (Linden, Muschalla, & Walter, Reference Linden, Muschalla and Walter2020). Also, the most frequently reported events may reflect to some point common psychotherapeutic processes (e.g. problem actuation) that are not inherently unwanted (Grawe, Reference Grawe1997). Thus, the presented treatment protocol can be regarded as safe.
Our results are in line with previous studies examining acceptability of CBT for PSS in other clinical populations, e.g. SSD, migraine, chronic fatigue (Adamson, Ali, Santhouse, Wessely, & Chalder, Reference Adamson, Ali, Santhouse, Wessely and Chalder2020; Klan, Liesering-Latta, Gaul, Martin, & Witthöft, Reference Klan, Liesering-Latta, Gaul, Martin and Witthöft2019; Sharpe et al., Reference Sharpe, Walker, Williams, Stone, Cavanagh, Murray and Carson2011; Verdurmen, Videler, Kamperman, Khasho, & van der Feltz-Cornelis, Reference Verdurmen, Videler, Kamperman, Khasho and van der Feltz-Cornelis2017). Developing interventions that are feasible and accepted by the target population seems important as such factors may positively influence therapeutic outcomes and the implementation of new treatment options under naturalistic conditions (Proctor et al., Reference Proctor, Landsverk, Aarons, Chambers, Glisson and Mittman2009). In fact, positive evaluations of general (e.g. alliance; Flückiger, Del Re, Wampold, & Horvath, Reference Flückiger, Del Re, Wampold and Horvath2018) and group-specific therapeutic processes (e.g. cohesion; Burlingame, McClendon, & Yang, Reference Burlingame, McClendon and Yang2018) have been shown to be positively related with the outcome of psychological interventions. For disorders that are marked by the presence of PSS, observational studies indicate that patients benefit more from biopsychosocial interventions when they are satisfied with treatment or perceive the intervention as helpful and appropriate for their symptoms (Beehler et al., Reference Beehler, Loughran, King, Dollar, Murphy, Kearney and Goldstein2021; Nees et al., Reference Nees, Riewe, Waschke, Schiltenwolf, Neubauer and Wang2020; Rief et al., Reference Rief, Bleichhardt and Timmer2002). In comparison, less is known about the influence of therapists’ evaluations of the therapeutic process. Similar to our results, therapists’ evaluations have been more critical compared with those of participants in previous studies (Puschner, Bauer, Kraft, & Kordy, Reference Puschner, Bauer, Kraft and Kordy2015; Viefhaus et al., Reference Viefhaus, Döpfner, Dachs, Goletz, Görtz-Dorten, Kinnen and Walter2019), but process-outcome relationships may be weaker for therapist ratings (Horvath & Symonds, Reference Horvath and Symonds1991). In general, health care providers may be more critical regarding delivered treatments due to their professional background, enabling them to reflect more in depth on an intervention. Upon study completion, we offered a feedback session to the study therapists to better understand the relatively lower satisfaction with the initial sessions and to identify potential needs for modification of the treatment protocol. Feedback was mostly related to a discrepancy between the time available and the proposed schedule of the sessions while generally being satisfied with content and interventions. We addressed the therapists’ feedback by shortening and further manualizing the schedule of the initial sessions in a modified version of the treatment protocol.
In addition, pre-post comparisons suggest that the treatment program may reduce subjective fatigue and improve disease coping on a cognitive level with small- to medium-sized within-group effects. This is consistent with previous studies showing that CBT effectively reduces fatigue severity in chronic fatigue syndrome and fatigue related to other diseases such as multiple sclerosis (Harrison et al., Reference Harrison, Safari, Mercer, Picariello, van der Linden, White and Norton2021; Price et al., Reference Price, Mitchell, Tidy and Hunot2008). Moreover, the observed changes in illness cognitions may be of particular relevance. In a recent meta-analysis, differences in pre-treatment cognitive characteristics have been shown to predict patients’ outcome in CBT for PSS suggesting that illness cognitions represent important features that should be addressed by psychological interventions (Sarter et al., Reference Sarter, Heider, Kirchner, Schenkel, Witthöft, Rief and Kleinstäuber2021). In line with that, CBT indirectly reduced fatigue severity in cancer patients by changing fatigue self-efficacy (Poort et al., Reference Poort, Müller, Bleijenberg, Verhagen, Verdam, Nieuwkerk and Knoop2021). Contrary to previous results (Kleinstäuber et al., Reference Kleinstäuber, Witthöft and Hiller2011), we did not find significant reductions of somatic and psychopathological symptom burden, but at the descriptive level differences tended in the expected direction. Interpretation of these results should consider a potentially inadequate treatment dose. Previous work has proposed a positive dose-response relationship for psychological interventions for somatic symptoms (Glombiewski et al., Reference Glombiewski, Sawyer, Gutermann, Koenig, Rief and Hofmann2010; Kleinstäuber et al., Reference Kleinstäuber, Witthöft and Hiller2011). In contrast, CBT-PCC comprised eight sessions delivered within four weeks. Considering the short period of time while simultaneously providing a broad variety of novel interventions, increasing the treatment dose to more sessions may be more helpful for participants to adopt the newly learned coping strategies. However, because CBT-PCC was embedded in routine inpatient rehabilitation and additional treatment components mostly complemented the cognitive-behavioral approach (e.g. physiotherapy, individual psychotherapeutic sessions, relaxation training), we assume the overall treatment dose the sample received as sufficient. Interestingly, for behavioral treatments of chronic pain, shorter treatments actually performed better than longer treatments, suggesting that individuals recover more quickly when offered fewer sessions from the outset (Glombiewski et al., Reference Glombiewski, Holzapfel, Riecke, Vlaeyen, de Jong, Lemmer and Rief2018). Another explanation may be the small time difference between pre- and post-intervention assessment. Despite evidence for a substantial improvement of disease coping in our sample, less is known about how long it takes for these changes to translate into a reduction of somatic or psychopathological symptom burden. Because the time difference between assessments was small and post-intervention assessment was administered before discharge from inpatient rehabilitation (i.e. after a maximum of four weeks and before return to daily life), the study design may not have been optimal to detect changes in these measures. Therefore, future studies should include longer follow-up periods to address this issue.
Participants in our sample experienced predominantly fatigue, but overall, reported symptoms were polysymptomatic and reported alongside elevated psychological distress. The average time since SARS-CoV-2 infection was approximately 62 weeks and the proportion of people reporting psychobehavioral features of somatic symptom distress was comparable to other cohorts with PCC after mild COVID-19 in a neurological setting (Fleischer et al., Reference Fleischer, Szepanowski, Tovar, Herchert, Dinse, Schweda and Kleinschnitz2022; Kachaner et al., Reference Kachaner, Lemogne, Dave, Ranque, de Broucker and Meppiel2022). In addition, a substantial proportion of our sample was diagnosed with a current mental disorder. The elevated psychopathological symptom burden among PCC patients, including the high rate of psychiatric comorbidities, corroborates the need for integrative approaches to the understanding and treatment of PCC (Lemogne et al., Reference Lemogne, Gouraud, Pitron and Ranque2023). Although we cannot elaborate on the temporal relationship between PCC and psychiatric comorbidities, their co-occurrence clearly demonstrates the relevance of psychotherapeutic treatment strategies for PCC. Besides CBT-PCC primarily targeting somatic symptom distress, many of the proposed interventions have a transdiagnostic character (e.g. stress management, cognitive restructuring), thus CBT-PCC might be particularly suited for dealing with psychiatric comorbidities. Consistent with this, our results demonstrated the acceptability of CBT-PCC for a chronically and complex burdened sample. Generalizability of our results to other phenotypes of PCC (e.g. post-intensive care syndrome; Yong & Liu, Reference Yong and Liu2022) may be restricted due to the exclusion of people with severe courses of acute COVID-19, but CBT has been shown to be indicated for PSS with varying extent of biological underpinnings (McNaughton et al., Reference McNaughton, Hush, Beath, Gandy, Dear and Jones2019). Nonetheless, research concerning the application of specific treatment options for different phenotypes of PCC is needed to allow for evidence-based, personalized treatment decisions.
Several limitations have to be considered when interpreting our results. The lack of a control group, the small sample size, and the relatively high drop-out rate restrict the interpretation of the data. More precisely, the study design does not allow treatment effects to be attributed to the intervention since we pursued a one-group design and participants additionally received routine inpatient treatment. In addition, health insurance coverage for inpatient rehabilitation did not cover the entire intervention duration in some cases, resulting in participants being discharged before study completion. Moreover, results may have been biased due to the self-selection of participants, and because discontinuation of the study was associated with baseline clinical characteristics. Lastly, information on the long-term effects of the intervention is missing.
Conclusion
The post-COVID-19 condition affects a significant proportion of people after COVID-19, leading to a substantial burden on both an individual and societal level (Davis et al., Reference Davis, Assaf, McCorkell, Wei, Low, Re'em and Akrami2021), but research on effective treatments for PCC is still at an early stage. Our results provide evidence for the feasibility, acceptability, and safety of a specialized CBT protocol in inpatients with PCC. With regard to treatment effects for PSS (Kleinstäuber et al., Reference Kleinstäuber, Witthöft and Hiller2011), we assume CBT-based interventions to be an adequate option for PCC as well, but so far, only few promising treatment protocols exist (Frisk et al., Reference Frisk, Jürgensen, Espehaug, Njøten, Søfteland, Aarli and Kvale2023; Kuut et al., Reference Kuut, Müller, Csorba, Braamse, Aldenkamp, Appelman and Knoop2023; Skilbeck, Reference Skilbeck2022). Therefore, future studies should examine the efficacy of CBT-PCC in randomized controlled trials. In addition, effective treatment options for other associated symptoms (e.g. cognitive deficits) are needed and should be combined within multidisciplinary rehabilitation approaches to address the complexity and diverse phenotypes of PCC.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291723002921.
Acknowledgements
The authors wish to thank Evelina Albach, Michaela Humbek, Sina Marchione, Isabella Oswald, and Katrin Wiesand for intervention delivery and ongoing feedback on the feasibility of the treatment protocol, Julian Merx with data acquisition, and Alan Ibrahim for help with data entry. The authors also thank the inpatients that consented to participate in the study.
Funding statement
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Competing interest
None.
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Open access
