Acceptance and commitment therapy for older people with treatment-resistant generalised anxiety disorder: the FACTOID feasibility study

What is generalised anxiety disorder?

Generalised anxiety disorder (GAD) is the most common anxiety disorder among older people, with estimated prevalence rates ranging from 1.2% to 11.2%. 1,2 The main symptoms that characterise GAD are excessive anxiety and worry, which the person experiences as difficult to control, as well as feelings of fear, dread and uneasiness that have occurred on more days than not for at least 6 months. Other symptoms include restlessness or feeling ‘on edge’, tiredness, irritability, muscle tension, difficulties with concentrating, difficulties with sleeping, shortness of breath, fast heartbeat, sweating and dizziness. 3 It is a condition that may persist for decades, with a mean symptom duration of 20–30 years, in older people across community, medical and mental health samples in multiple countries. 4,5

Generalised anxiety disorder in older people is associated with poorer health-related quality of life, increased disability, greater health-care utilisation, increased medication intake and functional limitations in comparison with non-anxious older people. 6–8 Comorbidity with other anxiety, mood and personality disorders is common and is associated with poorer outcomes. 6,9–12 For example, comorbid anxiety and depression is associated with more severe somatic symptoms, poorer social functioning, greater suicidal ideation and a higher likelihood of prescription of benzodiazepines, as well as poorer treatment response. 13,14 Several factors are associated with treatment-resistant anxiety, including comorbid physical and mental health conditions, noncompliance and environmental stressors. 13

How is generalised anxiety disorder currently managed in the NHS?

National Institute for Health and Care Excellence (NICE) guidelines currently recommend a stepped care approach to the management of GAD. 15 Step 1 comprises identification and assessment, followed by education and active monitoring within primary care. If symptoms have not improved, then in step 2 one or more low-intensity psychological interventions, such as guided self-help based on cognitive behavioural therapy (CBT) and psychoeducational groups, are offered, again within primary care. Should symptoms persist, or if there is marked functional impairment, then pharmacotherapy [(e.g. selective serotonin reuptake inhibitors) and/or high-intensity, individual psychotherapy (either CBT or applied relaxation)] are offered (step 3). Following this, if symptoms still persist, a referral to specialist mental health services (usually located within secondary care) for assessment and treatment is recommended as step 4. Suggested treatment options in step 4 include offering interventions from steps 1–3 that have been previously declined and offering combination therapy (e.g. pharmacological plus psychological therapy).

What is the evidence for the management of generalised anxiety disorder in older people?

With respect to psychological interventions, the majority of studies have examined the efficacy of CBT for GAD in older people. 8,16–26 Only a few have examined other psychological interventions such as acceptance and commitment therapy (ACT) and mindfulness-based stress reduction. 27,28 Pooled odds ratios in favour of these interventions when compared with waiting list or usual care controls have been reported, but not when compared with active controls or other forms of psychotherapy. 29

Furthermore, there is evidence of smaller treatment effect sizes among older people than among working age adults, as well as higher drop-out rates. 30–32 For example, a recent meta-analysis of CBT for GAD reported an overall effect size that was nearly double that achieved in younger people than in older people {g = 0.94 [95% confidence interval (CI) 0.52 to 1.36] in working-age adults vs. g = 0.55 (95% CI 0.22 to 0.88) in older people}. 33 As the authors of this meta-analysis noted, there is clearly room for improvement in the provision of effective psychological interventions for GAD in older people.

‘Treatment-resistant’ older people, that is, older people who fail to respond adequately to first-line pharmacological and psychological interventions, are by definition less likely to respond to treatment. Although there is no agreed definition of treatment-resistant generalised anxiety disorder (TR-GAD),34 when a person with GAD fails to respond to treatment after completing the first three steps of the stepped care approach,15 GAD can be considered to be resistant to treatment. A previous systematic review35 was unable to identify any randomised controlled trial (RCT) or prospective comparative study of either pharmacological or psychological interventions for treatment-resistant anxiety in older people. Given that the older adult population is projected to increase rapidly in the next 40 years,36 and hence more people will present with TR-GAD to older adult services, identifying effective interventions for this population is clearly a priority.

One possible intervention for managing TR-GAD in older people is CBT. However, as noted above, evidence of lower efficacy of CBT for GAD in older people than in working-age adults suggests that an alternative form of psychological intervention may be required. ACT could be a particularly promising candidate for this age group, given that older people with chronic pain respond better to ACT than to CBT, whereas younger people respond better to CBT than to ACT. 37 Consequently, the present study sought to investigate whether or not ACT is an acceptable and feasible approach for potentially managing TR-GAD in older people.

What is acceptance and commitment therapy?

Acceptance and commitment therapy is an acceptance-based behaviour therapy38 with a strong evidence base for improving outcomes (such as functioning, quality of life and mood) in chronic pain39 and a growing evidence base in chronic disease40 and mental health contexts. 41 It aims to (1) teach people new skills for managing thoughts, feelings and sensations; (2) help them to clarify what they value and what is important and meaningful to them in their lives; and (3) identify ways in which they can best live their lives in accordance with these values alongside the thoughts, feelings and sensations they may be experiencing. It achieves this through a variety of ‘core’ acceptance, mindfulness, commitment and behaviour change processes (as shown in Table 1), with the ultimate aim of increasing ‘psychological flexibility’. Psychological flexibility is defined as ‘the ability to contact the present moment more fully as a conscious human being and to either change behaviour or persist, when doing so serves valued ends’. 42 Research has supported the applicability of these core ACT processes in a variety of clinical populations, including in older people. 43

Acceptance and commitment therapy can be seen as a novel alternative to traditional forms of psychotherapy such as conventional CBT. The focus of conventional CBT is on alleviating distress or symptoms by changing how one thinks and behaves in emotional situations (e.g. by challenging the validity of negative thoughts or solving problems). The phrase ‘catch it, check it, change it’, in relation to negative thoughts, captures the essence of conventional CBT. By contrast, ACT is focused on increasing personally meaningful behaviour in the presence of distress and symptoms (although distress or symptoms may improve as a by-product of therapy). The phrase ‘Accept your experiences and be present, Choose a meaningful direction for your life, and Take action’ sums up ACT. 44

What is the rationale for acceptance and commitment therapy for older people with treatment-resistant generalised anxiety disorder?

There are several reasons why ACT may be a beneficial intervention for older people with TR-GAD. First, a preliminary RCT of ACT versus CBT for GAD in older people reported improvements in worry/anxiety and depression with both interventions and higher treatment completion rates with ACT than with CBT. 27 Although these results are promising, this study was limited by its small sample size (n = 7 in the ACT condition and n = 9 in the CBT condition) and limited applicability to older people with TR-GAD in the UK (as the study did not exclusively recruit those with TR-GAD and was conducted in the USA).

Second, the approach taken to managing unwanted thoughts, emotions and sensations in ACT may be a better fit for older people with TR-GAD than the approach taken in conventional CBT. ACT aims to reduce attempts to control, eliminate or avoid unwanted thoughts, emotions and sensations and to improve function through increased engagement in valued, meaningful activities. Conventional CBT, on the other hand, aims to change or suppress emotional experiences, for example by challenging the validity of unwanted thoughts or trying to eliminate or solve problems. Such approaches may not work well with older people with TR-GAD given that multiple, comorbid chronic physical and mental health conditions and multiple losses (e.g. to one’s health, family, social network, role/identity and financial status) are common in this population. This is because issues such as these may not be amenable to being solved or eliminated, and thoughts about them may be entirely valid. Furthermore, challenging the validity of worries about future losses may be perceived negatively by older people with TR-GAD because, although excessive and unhelpful, they may have an obvious basis in reality.

Third, there is evidence that control-orientated strategies, such as trying to eliminate problems that cannot be solved, are actually detrimental to older people’s well-being. 45 This may partly explain two related findings: first, why smaller effect sizes in favour of CBT for GAD, as well as higher drop-out rates, have been reported among older people than among younger people;30,32,33 and, second, why older people with chronic pain were more likely to clinically respond to ACT than CBT, whereas younger people were more likely to respond to CBT than ACT. 37 ACT, with its focus on increasing adaptive functioning and how best to live with such difficulties and worries (as opposed to challenging, changing or trying to eliminate them), may be more appropriate in this population. Supporting this, ACT has been shown to better fit the needs of people with disabling long-term conditions46 and may be particularly helpful when distress is associated with realistic or valid thoughts. 47

Finally, ACT has been found to be as effective as CBT and applied relaxation in the treatment of GAD in working-age adults. 48–51 Furthermore, greater recovery rates and lower drop-out rates have been reported with ACT than with CBT in the management of treatment-resistant mental health problems in working-age adults. 52 Whether or not ACT is similarly effective in older people with TR-GAD is clearly worthy of further investigation.

What is the evidence for acceptance and commitment therapy in older people with treatment-resistant generalised anxiety disorder?

Although ACT has been applied to a wide range of mental and physical health conditions including anxiety, depression and chronic pain,53,54 very few studies of ACT have been conducted with older people. The majority of studies have examined ACT for chronic pain,55–58 with only a few other studies focusing on GAD,27 veterans aged ≥ 65 years with depression59 and those living in long-term care facilities. 60 Beneficial effects on symptoms of depression, anxiety and functional measures have been reported in these studies, along with high rates of attendance. For example, 100% session attendance was reported in 7 out of 7 (100%) older people with GAD27 and in 59 out of 76 (78%) older people with depression. 59 However, to our knowledge, no studies to date have examined ACT specifically for older people with TR-GAD.

Aims

The aim of the current study was to develop an intervention based on ACT specifically for older people with TR-GAD, and to examine the feasibility and acceptability of its delivery in the NHS.

Objectives

The objectives of the current study were to:

• develop and refine a manualised intervention in accordance with Medical Research Council (MRC) guidelines for developing and evaluating complex interventions61 and using qualitative methodological approaches

• use qualitative interviews to explore the intervention’s acceptability and feasibility to older people with TR-GAD

• use a nationwide survey to clarify usual care for older people with TR-GAD (information that could be used for a future substantive trial)

• obtain quantitative and qualitative estimates of the acceptability and feasibility of the intervention and study methods in an open, uncontrolled, feasibility study

• clarify key study design parameters for a future substantive trial of clinical effectiveness and cost-effectiveness (e.g. the choice of comparator and outcome measures, and the number of recruitment sites based on referral/recruitment/attrition rates in the uncontrolled feasibility study).

Design

A person-centred approach was used to ground the development of the intervention in the perspectives and lives of the older people for whom it was intended. 63 Systematic, qualitative methods were used alongside patient and public involvement (PPI) to build on an ACT protocol previously piloted with a small number of older people with GAD (n = 7), but not specifically those with TR-GAD. 27 Stage 1 (intervention planning) investigated intervention preferences and priorities, relevant experiences, and barriers to and facilitators of engaging with talking therapy. Stage 2 (intervention design and development) involved formulating design objectives, and intervention features relevant to each objective, for the ACT intervention. Table 2 shows a summary of the person-based activities involved in each stage of intervention development.

Participants

Older people with TR-GAD were eligible to participate in the study if they met the following eligibility criteria.

Recruitment procedures

Procedure

Data analyses

We used the framework approach69 to facilitate analysis within and between individual cases and groups of participants. One author (KK) conducted the interviews and focus group with academic clinicians, listened to all recordings and repeatedly read the transcripts and research notes to familiarise herself with the data. We noted key issues, recurrent themes and interpretations, and discussed these in supervision and at research team meetings. Two additional authors (VL and RG), who had not analysed the other transcripts, reviewed three transcripts to help identify alternative viewpoints. We developed a descriptive theoretical framework of key beliefs about GAD, coping strategies and therapy specific to this group and considered relevant to the intervention by consensus and used this to index subsequent transcripts. Data were then charted into matrices to help map and interpret the data set as a whole; comparisons were made across themes and participants to help synthesise the findings.

Ethics

Ethics approval was granted by the London–Camberwell St Giles Research Ethics Committee (REC) on 9 May 2017 and Health Research Authority approval was granted on 12 May 2017 [Integrated Research Approval System (IRAS) identifier (ID) 214775, REC reference 17/LO/0704; see Report Supplementary Material 2].

Participants

As shown in Table 3, the majority of older people with TR-GAD were recruited from secondary care services (n = 8, 53%) and self-identified as female (n = 11, 73%), were in their 70s (n = 8, 53%), were married (n = 7, 47%) and were educated to at least degree level (n = 8, 53%). All self-identified as white/white British (n = 15, 100%). The majority of health-care professionals and academic clinicians self-identified as female (n = 27, 75%) and were most commonly clinical or counselling psychologists (n = 13, 36%) or psychiatrists (n = 10, 28%) working in secondary care settings (n = 25, 84%).

Key themes in stage 1

Interviews with older people and health-care professionals identified key issues, needs and challenges that would need consideration when developing the intervention. These were categorised into four key themes: (1) ‘expert in one’s own condition’, (2) ‘deep-seated coping strategies’, (3) ‘expert in therapy’ and (4) ‘support with implementation’. Subthemes within each key theme were also identified. We present data across the participant groups, with similarities and discrepancies highlighted where relevant. Sample quotations are presented in Appendix 1.

Guiding principles in stage 2

Themes identified in stage 1 were developed into guiding principles for therapy (Table 4), in consultation with the Service User Advisory Group, and modified in response to follow-up interviews with older people with TR-GAD and via further discussion with experts (clinical academics and therapists involved in the study). Some of the key features for optimising an ACT intervention for older people with TR-GAD are described in this section. The final outputs from the process of intervention design, development and optimisation have been presented rather than incremental changes being itemised.

How acceptance and commitment therapy was adapted for older people with treatment-resistant generalised anxiety disorder

A description of the specific adaptations made to the ACT intervention for older people with TR-GAD is presented below. This takes into account the guiding principles identified in Table 4, as well as previous recommendations with respect to using ACT with older people. 43

Design

This was a cross-sectional study.

Participants

Participants were:

• Older people aged ≥ 60 years who were experiencing or had experienced difficulties with GAD or worrying. All older people who self-identified as experiencing difficulties with worrying were invited to complete the survey, as opposed to just those who were experiencing TR-GAD, because it was thought that participants might find it difficult to identify whether or not they qualified as ‘treatment resistant’ (i.e. failed to respond after completion of steps 1–3 of the stepped-care approach for GAD).

• Health-care professionals who work with this population of older people in primary and secondary care settings (including GPs, psychiatrists, clinical psychologists, community psychiatric nurses, occupational therapists and social workers).

Settings

Older people were recruited from the community via convenience sampling through GP practices (via local Clinical Research Networks across the UK), primary and secondary care services (IAPT services and CMHTs for older people), a mental health charity [Mind, URL: www.mind.org.uk (accessed 2 February 2021)], local community groups [University of the Third Age (U3A) groups], a service user research forum [INVOLVE, URL: www.invo.org.uk/communities/information-for-members-of-the-public/ (accessed 2 February 2021)], and online forums for older people [Fifty Plus Forum, URL: www.fiftyplusforum.co.uk (accessed 2 February 2021); Pensioners Forum, URL: www.pensionersforum.co.uk/ (accessed 2 February 2021); Senior Forums, URL: www.seniorforums.com (accessed 2 February 2021); Buzz 50, URL: www.buzz50.com (accessed 2 February 2021); Senior Chatters, URL: https://seniorchatters.co.uk (accessed 2 February 2021)]. Several other online forums for older people were approached to assist with advertising the survey but declined involvement or did not permit this [Age UK, URL: www.ageuk.org.uk/get-involved/social-groups/older-peoples-forums/ (accessed 2 February 2021); Gransnet, URL: www.gransnet.com/forums (accessed 2 February 2021); Silver Surfers, URL: www.silversurfers.com/silversurfers-forum/ (accessed 2 February 2021)].

Health-care professionals were identified through GP practices (via local Clinical Research Networks across the UK), online forums associated with occupation-specific organisations (Royal College of Psychiatrists, Royal College of Nursing, The British Psychological Society) and an online forum for those working in health and social care [Contact, Help, Advice and Information Network (CHAIN), URL: www.chain-network.org.uk (accessed 2 February 2021)]. The Royal College of Occupational Therapists was approached to assist with advertising the survey but declined involvement.

Survey

The brief online survey comprised a series of multiple-choice questions with free-text boxes that enabled provision of further information if desired. There were two versions of the survey – one for service users and one for health-care professionals – because the content of questions and terminology differed for the two groups of respondents. The survey was kept as brief as possible to maximise completion rates and took approximately 5–10 minutes to complete.

Procedure

We contacted a range of organisations across the UK and asked them to advertise the survey (see Settings). The survey used a web-based survey tool, Opinio, version 7 (2017) (ObjectPlanet, Inc., Oslo, Norway). A paper-based version of the survey was available on request but was not routinely distributed as originally hoped owing to resource limitations (both finances and time). A participant information sheet was provided at the beginning of the survey; if participants proceeded to complete the survey, it was assumed that they were providing their consent to participate. The survey was open for data collection from September 2017 to December 2017. The survey was piloted with our PPI group prior to data collection. All responses were anonymous.

Rural Urban Classification83 was used to categorise participants’ geographical area of residence (based on postcode or closest city/town). Areas with a population > 10,000 were categorised as urban and those with a population ≤ 10,000 were categorised as rural.

Data analyses

Responses were excluded if participants indicated that they were aged < 60 years or had ‘never experienced difficulties with long-term worrying or their nerves’. Those with current symptoms of GAD and a treatment history suggestive of TR-GAD were identified in post hoc analyses so that what constitutes usual care in this specific group of respondents could be clarified. Current symptoms of GAD were defined as a score of > 5 points on the GAD-7 and a treatment history suggestive of TR-GAD was defined as either having received at least two types of treatment (pharmacotherapy and/or psychological therapy) for worrying or having been offered at least two types of treatment (pharmacotherapy and/or psychological therapy) for worrying and having refused.

Data relating to demographic and clinical characteristics of service users and demographic and professional characteristics of health-care professionals were summarised using frequencies and percentages, and means and standard deviations (SDs) or medians and interquartile ranges (IQRs) for very skewed distributions (i.e. skewness values < –2 or > 2). 84 Perceived helpfulness of treatments was rated on a 5-point Likert scale from 1 (not at all helpful) to 5 (extremely helpful) and was summarised using medians and IQRs.

Parametric rather than non-parametric tests were chosen to analyse Likert scale responses, as previously recommended. 85,86 Data pertaining to the perceived helpfulness of treatments in service users and health-care professionals were analysed as follows. The perceived helpfulness of treatments in service users was examined by submitting data to an exploratory two-way (treatment × time) within-subjects analysis of variance (ANOVA), with type of treatment (pharmacotherapy vs. psychological therapy vs. a combination of pharmacotherapy and psychological therapy) and time (current vs. past) as within-subjects variables. The perceived helpfulness of treatments in health-care professionals was examined by submitting data to a one-way within-subjects ANOVA, with type of treatment as a within-subjects variable. Post hoc Bonferroni-adjusted pairwise comparisons were used to examine differences between conditions, where appropriate.

Ethics

Ethics approval was granted by the London–Camberwell St Giles Research Ethics Committee on 9 May 2017, with Health Research Authority approval granted on 12 May 2017 (IRAS ID 214775; REC reference 17/LO/0704).

Service users

A total of 136 service users completed the online survey. Responses were excluded for three service users (2%) who were aged < 60 years and for 11 service users (8%) who had ‘never experienced difficulties with long-term worrying or their nerves’. Therefore, we analysed data from 122 service users (90%). Post hoc analyses identified 58 service users (48%) who were considered to have current symptoms of GAD and a treatment history suggestive of TR-GAD. We report data for all service users in Appendix 2.

As shown in Table 5, the majority of service users were aged 65–74 years (n = 33, 57%). Most self-identified as female (n = 55, 95%) and white/white British (n = 54, 93%) and resided in urban areas (n = 50, 86%), mainly in the London region (n = 34, 59%). Just under half of service users had a degree or postgraduate qualification (n = 28, 48%), with only five service users (9%) reporting no educational qualifications.

With respect to clinical characteristics displayed in Table 6, there was a wide variation in the reported number of years experiencing difficulties with long-term worrying, with ≥ 30 years (n = 25, 43%) being the most common. The overall mean total score on the GAD-7 was in the moderate range (14.0 points, SD 4.3 points), and 75% of service users scored in the moderate to severe range. The mean number of comorbid mental health complaints was 4.1 (SD 1.9): comorbid symptoms of depression (n = 53, 91%), panic (n = 40, 69%), post-traumatic stress disorder (PTSD) (n = 36, 62%), social phobia (n = 35, 60%) and agoraphobia (n = 30, 52%) were the most commonly reported.

In terms of reported treatments for worrying in the past or currently (either received or refused), just under half of service users were currently receiving pharmacotherapy (n = 28, 48%), with antidepressants being the most commonly reported medication (n = 28, 62%; Table 7). In contrast, only one-quarter of service users (n = 13, 24%) were currently receiving psychological therapy, with counselling being most commonly reported (n = 4, 44%). Most service users had tried some form of pharmacotherapy or psychological therapy in the past (n = 42, 72%, and n = 45, 83%, respectively). One-quarter of participants (n = 14, 26%) reported trying at least three different types of pharmacotherapy in the past, and just under half reported trying at least two different types of psychological therapy in the past (n = 26, 48%). The most common forms of pharmacotherapy tried in the past were antidepressants (n = 24, 45%) and sedatives (n = 19, 36%), whereas the most common forms of psychological therapy were CBT (n = 20, 40%) and counselling (n = 9, 18%). Approximately one-third of service users reported refusing pharmacotherapy in the past (n = 17, 30%), whereas the number of service users who reported refusing psychological therapy in the past was small (n = 4, 8%). Forty-two per cent of service users reported not receiving any form of current pharmacotherapy or psychological therapy, despite the mean GAD-7 total score being in the moderate range (13.3 points, SD 4.0 points) for this subsample, with five service users scoring in the mild range, 10 scoring in the moderate range and eight scoring in the severe range.

Service users rated the perceived helpfulness of current and past treatments for worrying on a scale from 1 (not at all helpful) to 5 (extremely helpful). As shown in Table 8, there was little variation in the median ratings across current and past treatments, although an inspection of the data suggested that there was a possible trend for medication alone to be perceived as less helpful than other forms of treatment.

Data were submitted to ANOVAs to compare ratings across different types of treatment. Unfortunately, it was not possible to conduct a two-way ANOVA with time (current vs. past) and treatment (pharmacotherapy vs. psychological therapy vs. a combination of pharmacotherapy and psychological therapy) as within-subjects variables because data were available for only 11 service users owing to missing data or no ratings being given (i.e. a ‘not applicable’ response). Consequently, separate one-way ANOVAs for current and past treatment were conducted, with treatment (pharmacotherapy vs. psychological therapy vs. a combination of pharmacotherapy and psychological therapy) as a within-subjects variable. The main effect of current treatment [F(2,20) = 1.14, mean squared error (MSE) = 0.27, p = 0.340, η_p² = 0.102] and the main effect of past treatment [F(2,52) = 1.57, MSE = 0.78, p = 0.218, η_p² = 0.057] were non-significant. There were no statistically significant differences in perceived helpfulness across different types of treatment, either currently or in the past (Table 9).

Health-care professionals

Data from 86 health-care professionals who completed the online survey were analysed. As shown in Table 10, approximately two-thirds of health-care professionals were in their 40s or 50s (n = 59, 69%). The majority self-identified as female (n = 65, 76%) and white/white British (n = 74, 86%) and worked in urban areas (n = 76, 89%), primarily in the South West, South East and London regions of England (n = 64, 74%). The most common clinical professions were psychology (n = 27, 32%) and psychiatry (n = 24, 28%), and medicine for older people; social work and occupational therapy were the least represented professions (total n = 4, 5%). There was wide variation in the level of profession, although 60% of health-care professionals (n = 49) were in senior positions (e.g. GP, consultant, grade 8). Correspondingly, 51% of health-care professionals (n = 43) had been qualified for ≥ 16 years. The majority worked in secondary care (n = 58, 70%) and in the NHS (n = 76, 91%).

Table 11 presents data on the reported treatment of older people with TR-GAD. Health-care professionals estimated that, as a percentage of all people seen per month, an average of just over three-quarters (78.7, SD 29.1) were aged ≥ 65. However, among these, health-care professionals estimated that only one-fifth had TR-GAD or clinically significant chronic anxiety (20.6, SD 20.1). In terms of the types of treatment typically offered or referred for in this population, the most common types of pharmacotherapy were antidepressants (n = 55, 64%) and anti-epileptics (n = 26, 30%), with CBT (n = 66, 77%), mindfulness-based therapy (n = 46, 54%) and relaxation therapy (n = 42, 49%) being the most common types of psychological therapy. Just over half of health-care professionals reported typically offering or referring for pharmacotherapy ‘a lot of the time’ or ‘always’ (n = 40, 52%), which rose to 65% (n = 50) for psychological therapy. The most commonly endorsed reasons for not offering or referring for pharmacotherapy were patient preference (n = 53, 62%), side effects (n = 40, 47%) and comorbid health problems (n = 39, 45%). Patient preference was the most common reason for not offering or referring for psychological therapy (n = 52, 61%), followed by a lack of availability of this type of treatment (n = 30, 35%).

Ratings of the perceived helpfulness of treatments for older people with TR-GAD or clinically significant chronic anxiety by health-care professionals are shown in Table 12. There was some variation in the median ratings across pharmacotherapy and psychological therapy, with a possible trend for the majority of pharmacotherapy to be perceived as less helpful than psychological therapy. Data were submitted to a one-way ANOVA, with type of treatment as a within-subjects variable, to statistically compare ratings across different types of treatment.

It was possible to include data from only 27 health-care professionals owing to missing data or no ratings being given (i.e. an ‘I have no opinion’ response). Mauchly’s test of sphericity was significant (p < 0.05) and, therefore, Greenhouse–Geiser-corrected degrees of freedom and p-values are reported. There was a statistically significant main effect of the type of treatment [F(5.6,146.4) = 22.39, MSE = 23.73, p < 0.0005, η_p² = 0.46]. Post hoc Bonferroni-adjusted pairwise comparisons showed that CBT, mindfulness-based therapy and relaxation therapy were rated as more helpful than all medications excluding antidepressants (Table 13). In addition, CBT was rated as more helpful than counselling. Antidepressants were rated as more helpful than all other medications excluding anxiolytics. Medications with potentially sedative effects (sedatives, hypnotics and antihistamines) were rated the least helpful of all treatments.

Main findings with respect to usual care

Approximately half of service users (n = 28, 48%) reported currently receiving pharmacotherapy, with antidepressants being the most commonly reported (n = 28). By contrast, only one-quarter of service users (n = 13, 24%) reported currently receiving psychological therapy, with counselling being most commonly reported (n = 4, 31%). A substantial proportion of service users (n = 23, 42%) reported not receiving any form of current pharmacotherapy or psychological therapy, despite the mean GAD-7 total score being in the moderate range (13.3 points, SD 4.0 points) for this subsample.

To be included in these analyses, service users had to have received or been offered at least two types of treatment (pharmacotherapy and/or psychological therapy) for worrying in the past. Again, the most common types of pharmacotherapy were antidepressants (n = 24) and sedatives (e.g. benzodiazepines, n = 19), whereas the most common forms of psychological therapy were CBT (n = 20) and counselling (n = 9). Approximately one-third of service users reported refusing pharmacotherapy in the past (n = 17, 30%), whereas the number of service users who reported refusing psychological therapy in the past was small (n = 4, 8%).

Health-care professionals estimated that, on average, only one-fifth of older service users (i.e. those aged ≥ 65 years) whom they saw each month were experiencing TR-GAD or clinically significant chronic anxiety. They reported that the most common types of pharmacotherapy they offered or referred for were antidepressants (n = 55, 64%) and anti-epileptics (n = 26, 30%). CBT, mindfulness-based therapy and relaxation therapy were the most common types of psychological therapy that they offered or referred for (n = 66, 77%; n = 46, 54%; and n = 42, 49%, respectively). Patient preference was the most common reason for not offering or referring for pharmacotherapy or psychological therapy (n = 53, 62%; and n = 52, 61%, respectively).

Service users rated different types of pharmacotherapy and psychological therapy as equally helpful, both currently and in the past, with mean ratings in the slightly to moderately helpful range. By contrast, health-care professionals rated CBT, mindfulness-based therapy and relaxation therapy as more helpful than all medications excluding antidepressants, with mean ratings in the moderately helpful range for these forms of psychotherapy. This is consistent with the types of treatment that health-care professionals reported most frequently offering or referring for. Medications with potentially sedative effects (i.e. sedatives, hypnotics and antihistamines) were rated the least helpful of all treatments, with mean scores in the not at all helpful to slightly helpful range.

Research implications

There are a number of implications with respect to study design that arise from the survey findings. The first relates to the choice of comparator group for a future substantive trial of ACT. There are three types of comparators that could be used in such a trial: a non-active control arm (such as treatment as usual), an active control arm (such as a supportive talking control) or an alternative treatment. Non-active controls do not control for non-specific factors, such as therapist time and social attention, whereas active controls do. However, active controls do not permit comparisons with what the NHS is currently offering, given that active controls are not usually part of treatment as usual.

If a non-active control (such as treatment as usual) was used then it might be expected that one-quarter of participants in the treatment-as-usual arm would receive psychological therapy, which might dilute any treatment effect. On the plus side, the results of the survey suggest that very few would be offered or referred for ACT. However, the fact that treatment as usual does not typically include access to psychological therapy might lead to higher withdrawal rates in a non-active control arm owing to resentful demoralisation (i.e. participants dropping out because they wanted to be allocated to the intervention arm). One way of overcoming this might be to offer access to study materials after a final follow-up outcome assessment. Possibly in opposition to the notion of resentful demoralisation is the observation that some participants in our survey reported refusing pharmacotherapy or psychological therapy in the past (n = 17, 30%; and n = 4, 8%, respectively). Participants such as these may actually prefer to be in a non-active control arm, although whether or not they would even agree to be involved in a treatment trial in the first place is, of course, debatable.

On the other hand, if an active control (such as a supportive talking control) was used then this not only might serve as an incentive to participate in the trial but might also reduce the chances of withdrawal due to resentful demoralisation. Supporting this, no differences in drop-out rate were found between CBT and talking–treatment control arms in a RCT of CBT for older people with depression. 87 An additional bonus of this would be the increased scientific rigour that comes with controlling for non-specific factors such as therapist time and social attention. However, research costs would be increased because provision of a supportive talking control would constitute a research activity rather than an excess treatment cost. Furthermore, a larger sample size would be required to detect a treatment effect, and such a trial would not be able to answer the question of whether or not the intervention was better than treatment as usual.

Of course, one way of overcoming these issues might be to include both a non-active control (such as treatment as usual) and an active control (such as a supportive talking control) in a future substantive trial of ACT. This study design was used in a RCT of CBT for older people with depression. 87 However, this would further increase research costs, and necessitate an even larger sample size to randomly allocate a sufficient number of participants to each arm. An alternative to this would be to conduct a non-inferiority RCT in which the equivalence of ACT to current front-line treatment would be investigated. The feasibility of such an approach has been demonstrated in a non-inferiority RCT of ACT compared with CBT in working-age people with mixed anxiety disorders. 88 The results of our survey with respect to perceived helpfulness of treatments suggest that an appropriate comparison condition would be psychological therapy (such as CBT) rather than pharmacotherapy.

The second implication of the survey findings relates to possible inclusion criteria in a future substantive trial. The necessity of a medication dose stabilisation period (e.g. 4 weeks) is highlighted by the fact that approximately half of service users in this study were currently receiving pharmacotherapy. Furthermore, the fact that only one-quarter of service users reported currently receiving psychological therapy suggests that the standard procedure of asking participants to refrain from receiving other concurrent formal psychological therapy during the receipt of ACT would not be problematic in this population. However, it should be noted that these findings may be unrepresentative and may simply reflect the type of participants who completed this online survey (e.g. those who are interested in actively pursuing treatment for TR-GAD).

The third implication of the survey findings relates to recruitment rates. Health-care professionals estimated that, on average, only one-fifth of older service users (i.e. aged ≥ 65 years) who they saw each month were experiencing TR-GAD or clinically significant chronic anxiety. Health-care professionals reported working in a range of services in primary and secondary care settings, with the majority working in secondary care (n = 58, 70%). It might be expected that those working in secondary care would tend to see more older people with TR-GAD or clinically significant chronic anxiety than those in other settings. However, even in this subgroup of health-care professionals, 48% estimated that they were seeing only 0–10% of older service users with TR-GAD or clinically significant chronic anxiety each month, with 36% estimating this to be 11–30%. Such low identification rates may be expected given that GAD is a frequently undiagnosed condition in clinical practice, often being misdiagnosed as major depressive disorder with comorbid anxiety or ‘mixed anxiety and depression’ rather than GAD. 2,89 This suggests that referring clinicians in a future RCT may need additional support to help identify potential participants with TR-GAD on their caseloads.

Finally, the fact that the survey was not able to examine usual care for older people with TR-GAD across the UK as well as we had hoped suggests that an online survey might be useful in the site identification/set-up process of a future substantive trial. This could be used to (1) ascertain whether or not usual care provision differs across Clinical Commissioning Groups, (2) assess the degree to which access to ACT is routinely available at specific sites and (3) explore whether or not any routinely available services might conflict with eligibility criteria.

Strengths and limitations

This is the first study, to our knowledge, to explore the usual care that is offered to and received by older people with current symptoms of GAD and a treatment history suggestive of TR-GAD in the UK. It would appear that usual care is quite limited, focusing mainly on pharmacotherapy, and, on average, is perceived as being only slightly to moderately helpful. Therefore, it provides some indication of what usual care might look like in a larger-scale RCT of ACT for older people with TR-GAD. Furthermore, care was taken to ensure that the survey used terminology that older people with TR-GAD could understand and identify with. For example, the term ‘difficulties with long-term worrying or your nerves’ was used rather than ‘generalised anxiety disorder’. This was deemed necessary given that GAD is a frequently undiagnosed condition in clinical practice, as noted in Research implications, and many may not have identified with the formal diagnostic label.

However, there are a number of limitations in the current study, which mean that findings should be interpreted with caution. First, the majority of service users self-identified as female (n = 55, 95%) and white/white British (n = 54, 93%) and resided in urban areas (n = 50, 86%) in England, primarily London (n = 34, 59%). Being female is one of the known risk factors for anxiety disorders in older people90 and, therefore, a higher proportion of female service users is to be expected. Similarly, a higher proportion of people living in urban locations is to be expected given that 83% of the UK population live in urban locations. 91 However, it does mean that findings cannot be generalised to men, those from ethnic minorities or those living in rural locations or geographical locations other than London. Furthermore, it may not be possible to generalise findings to service users who do not have access to the internet or a personal computer (PC) given that data were collected via an online survey. Finally, as noted in Research implications, these findings may be representative of the type of participants who completed this online survey (e.g. those who are interested in actively pursuing treatment for TR-GAD). This suggests that a larger-scale investigation of usual care in men, those living in rural settings, those living in geographical locations other than London, those from ethnic minorities and those without access to the internet or a PC is warranted. A similar issue with representativeness was seen with health-care professionals, with the majority self-identifying as female (n = 65, 76%) and white/white British (n = 74, 86%), in their 40s and 50s (n = 59, 69%) and working in urban areas (n = 76, 89%) in southern regions of England (n = 64, 74%). This supports the notion that a larger-scale investigation of usual care offered by health-care professionals and received by service users in more diverse populations is warranted.

Second, the sample size with respect to those with current symptoms of GAD and a treatment history suggestive of TR-GAD (n = 58, 48% of the total data set) and health-care professionals (n = 86) was relatively small, which may have increased the likelihood of a type II error (i.e. reporting a false-negative finding). Appropriately resourcing the study to enable recruitment via GP surgeries and CMHTs across the UK and to enable the provision of a postal survey may have increased recruitment rates. Contacting Clinical Commissioning Groups across the UK may have also increased recruitment rates of health-care professionals.

Third, it was not possible to conduct a formal assessment of GAD, nor was it possible to verify the accuracy of self-reported information with respect to treatment history, etc. Furthermore, single-item screening questions were used to assess the presence of comorbid mental health conditions. Thus, some service users may not have actually been experiencing TR-GAD and some information may have been inaccurate, which may have contaminated the findings. Some confidence that service users were experiencing TR-GAD can be gained from the fact that nearly half of service users (43%) reported difficulties with long-term worrying for ≥ 30 years, 75% of service users scored in the moderate to severe range on the GAD-7 and the mean number of comorbid mental health complaints was 4.1, with symptoms of depression being the most common (91%). This fits what is known about the clinical picture of GAD in older people. For example, psychiatric comorbidity is known to be common in GAD,2 and the co-occurrence of depression and GAD is associated with greater chronicity92 as well as poorer treatment responsiveness and higher severity of GAD among older people. 93

Finally, the use of subjective rather than objective ratings of the helpfulness of treatments for worrying may have introduced a degree of variability in responses from both service users and health-care professionals. This is because subjective interpretations of ‘helpfulness’ may differ across individuals. Defining what was meant by ‘helpfulness’ may have helped to reduce variability in responses.

Design

This was an open, uncontrolled feasibility study that employed both quantitative and qualitative methodological approaches.

Participants

Participants were included in the study if:

• They were aged ≥ 65 years with a primary diagnosis of GAD as determined by the Mini-International Neuropsychiatric Interview (MINI)97 and the Structured Clinical Interview for DSM-IV Axis II Disorders. 65

• They had failed to respond to treatment in steps 1–3 of the stepped-care approach for GAD (i.e. 6 weeks of an age-appropriate dose of antidepressant medication or a course of individual psychotherapy), failed to tolerate this treatment or had previously refused this treatment and were still symptomatic. When determining whether or not a person had failed to respond to treatment for GAD, if they had remitted and then relapsed in relation to GAD then treatment received prior to remission was not considered when deciding whether or not they currently met criteria for treatment resistance.

• They were living in the community.

• They were able to provide informed, written consent.

• They had sufficient understanding of English to enable them to engage in ACT and complete patient-reported outcome measures.

• They had not participated in phase 1 of the study (i.e. qualitative interviews focused on development of the intervention).

• There was a minimum of a 1-month interval between previous psychotherapy and engagement in ACT.

Participants were not able to participate in the study if:

• they had a diagnosis of dementia

• they had a total score of < 25 points on the SMMSE66

• they were currently receiving ongoing psychotherapy or were unwilling to refrain from engaging in other forms of psychotherapy during the receipt of ACT

• they were expressing suicidal ideation with active intent (because an inpatient admission would have been more appropriate management)

• there were other medical or psychosocial factors that could compromise full study participation, such as imminently life-limiting illness or severe sensory deficits (e.g. blindness)

• they were diagnosed with intellectual disabilities.

It is common to specify a psychotropic drug stabilisation period (e.g. a stable dose for at least 2 months) as one of the inclusion criteria in psychotherapy studies to control for the potential confound of pharmacotherapy on mental well-being. However, this was not included in the current study as it was thought that this would have a significantly negative impact on recruitment, as many potential participants would be either ineligible or unwilling to wait for drug stabilisation to occur before receiving psychotherapy. Service users are frequently referred to CMHTs for medication reviews, which usually entails switching to another psychotropic drug, augmenting current treatment with another psychotropic drug or changing psychotropic drug dosages, and so it can take a number of months before a stable dose is achieved. Instead of including a psychotropic drug stabilisation period, all drug use was monitored during the course of therapy and accounted for in subsequent data analyses, where necessary.

Setting

Potential participants were recruited from a number of sources:

• referrals from organisations within primary care settings, including GP surgeries in the North Thames Clinical Research Network, South London Clinical Research Network and IAPTs

• referrals from secondary care settings, including CMHTs for older people in Camden and Islington NHS Foundation Trust; South London and Maudsley NHS Foundation Trust; Barnet, Enfield and Haringey Mental Health NHS Trust; and North East London NHS Foundation Trust

• self-referral from the community.

Recruitment procedures

Potentially eligible participants from primary and secondary care settings were identified and approached about the study in one of four ways. First, clinicians from GP surgeries, IAPTs and CMHTs identified and approached potentially eligible participants and sought verbal consent for members of the research team to contact them. Older people who meet diagnostic criteria for GAD are frequently referred to IAPTs and CMHTs with a diagnosis of major depression and comorbid anxiety rather than GAD. Consequently, clinicians in these services were asked to screen service users who were referred with a diagnosis of major depression and comorbid anxiety for GAD using the GAD-7. 82 For any service user scoring ≥ 11 on this scale (corresponding to moderate or severe GAD), clinicians asked the service user which symptoms were most distressing, severe or of most concern to them. If symptoms of GAD were most distressing, most severe or of most concern to them, or if symptoms of GAD and depression were equally problematic, then the clinician discussed the study with them and sought verbal consent for members of the research team to contact them. If symptoms of depression were most distressing, most severe or of most concern to them then the clinician referred them for appropriate treatment.

Second, potentially eligible participants were identified and approached through local ‘consent for research contact’ procedures. Any potentially eligible participants who had previously agreed to being contacted about research were identified by clinicians and then contacted by band 5 clinicians.

Third, nurses or GPs in participating GP practices (i.e. those who had expressed an interest in participating in research through primary care research networks and local contacts) conducted searches of their electronic medical records to identify those with diagnoses of GAD and other chronic anxiety states using a broad list of Read codes, a hierarchical coding system used to record clinical information such as diagnoses. 98 Once potentially eligible participants were identified through electronic medical record searches, nurses or GPs in participating GP practices approached them and sought verbal consent for members of the research team to contact them.

Finally, leaflets and advertisements were distributed in GP surgeries and other community settings, such as luncheon clubs and activity groups for older people, to facilitate self-referral. Leaflets included the GAD-7,82 a brief seven-item screening tool, and potential participants were invited to contact the research team if they scored ≥ 11 points on this scale (corresponding to moderate or severe GAD and above). In addition, the study was promoted through presentations at local community groups (e.g. luncheon clubs and activity groups) and team meetings in IAPTs and CMHTs.

Once potentially eligible participants were identified and verbal consent for contact was obtained, eligible participants were contacted by telephone by a member of the research team. The study was described to them and any questions or concerns about it were discussed. If they expressed an interest in participating in the study then, with their verbal consent, they were sent an information sheet and invitation letter. They were then contacted a minimum of 48 hours later by a member of the research team to determine whether or not they were still interested in participating in the study. If they were, then an appointment for a screening interview (lasting 1.5 hours) was arranged with a member of the research team. Written informed consent was sought from potentially eligible participants by a band 5 clinician, research nurse or research assistant during this screening interview, after which eligibility for inclusion in the study was determined.

Intervention

The intervention comprised up to 16 individual (i.e. one-to-one) sessions of ACT, delivered in person, with each session lasting up to 1 hour. This was consistent with previous recommendations of 12–16 sessions of ACT for older people,43,59 as well as with step 3 of the stepped-care approach for GAD. 15 Step 3 states that CBT or applied relaxation should usually consist of 12–15 sessions, with more sessions being offered if clinically required. The provision of 16 sessions allowed therapists to work at a slower pace, which is a recommended compensatory strategy for age-related cognitive changes in memory, attention and processing speed when working with older people. 78 There was a phased ending to the sessions, such that they were weekly for the first 14 sessions and then fortnightly thereafter, as some older people experience difficulties when therapy ends abruptly. An individual, rather than group, therapy approach was used because of the anticipated complexity of each individual’s presenting problems. The intervention was designed so that it could be delivered in the GP surgery, outpatient clinic or service user’s home (for those who were unable to travel because of physical and/or mental health issues). The location of the intervention delivery depended on participant preference and service restrictions. The intervention was tailored to the needs of older people with TR-GAD following the findings of qualitative interviews outlined in Chapter 2, as well as previous suggestions. 43 Table 14 and Figure 1, and Table 15 and Figure 2, list the outline of sessions, together with accompanying home practice tasks. A therapist manual and accompanying client workbook were developed to support the delivery of the intervention (available on request from the corresponding author).

FIGURE 1.

An outline of sessions 6–15, together with associated ACT metaphors and exercises. CWTPM, contact with the present moment.

FIGURE 2.

An outline of the home practice for sessions 6–15, together with associated ACT metaphors and exercises. CWTPM, contact with the present moment.

As can be seen in Table 14, the intervention addressed the six core ACT processes, in addition to exploring the workability of current strategies used to avoid, change or get rid of anxiety and worry. Each session was associated with a specific set of skills, metaphors, experiential exercises and home practice tasks designed to increase psychological flexibility. All sessions, bar the first and last sessions, followed the same structure. Sessions commenced with a short mindfulness exercise that was designed to increase awareness of the present moment, because some older people may experience difficulties in recognising, describing, observing or being aware of their internal states. 43 This was followed by brief ratings of worry, the struggle with worry and the degree to which the participant had been living a rich, full and meaningful life in the past week to help monitor progress. The presence of suicidal ideation (as well as any plans, intent and protective factors, if necessary) was also assessed at this point. Following this, there was a recap of the concepts and issues discussed in the previous session, as well as a discussion of the participant’s experience of the home practice. The rest of the session was spent addressing the target ACT process for that week’s session. The session ended with a summary of what had been discussed in the session, as well as a discussion of that week’s home practice.

The intervention was manualised, with the order and content being pre-determined for sessions 1–5 and the content alone being pre-determined for sessions 6–16. Consistent with the emphasis on psychological flexibility in ACT interventions, therapists had flexibility in deciding the order of delivery of sessions 6–16. In addition, they had a choice about which and how many metaphors and experiential exercises would be delivered in each session. For example, therapists may have been given the choice of three experiential exercises in a session, and may have been asked to deliver one or two of them. This meant that the pace of the sessions could be modified, depending on the needs and abilities of the individual client. Therapists also had flexibility in deciding the content of the ‘Other issues’ session, depending on the client’s needs and goals. The focus of this session could be on (1) withdrawing from prescribed or illicit substances (e.g. benzodiazepines); (2) problem-solving for external problems (as opposed to internal problems such as difficult thoughts, feelings and sensations); (3) establishing good sleep habits; (4) graded exposure; or (5) repeating a previous session (e.g. a session from which it was felt that the client would benefit from further practice). Overall, having some aspects of the sessions that were predetermined and others that were not gave therapists sufficient room to respond flexibly to clients’ idiosyncratic presentations while helping them to feel sufficiently supported to deliver the intervention. Feeling sufficiently supported was particularly important as the majority of therapists were new to ACT, as detailed below.

Training

Therapists initially attended a 4-day experientially based ACT training workshop, which was developed and delivered by members of the research team with training and experience in delivering ACT (MS and RG). There is no formal training pathway for ACT in the UK, so ACT is typically learnt through attendance at training workshops delivered by experts in the field and supervision of practice. Marc A Serfaty is a consultant psychiatrist and British Association for Behavioural and Cognitive Psychotherapies-accredited CBT therapist (1994) and fellow (2016). He had attended 12 days of ACT training workshops and had > 12 years’ experience in weekly ACT supervision by the time of training. In addition, he had 2 years’ experience of co-ordinating weekly ACT peer group workshops, as well as previous experience conducting a RCT of ACT for advanced cancer. Rebecca L Gould is a clinical psychologist with 9 years’ post-qualification experience of delivering psychotherapy interventions. She had attended 10 days of ACT training workshops and had 3 years’ experience in delivering ACT by the time of training. In addition, she had completed level 1 mindfulness teacher training through Bangor University, which shares some features with ACT.

The initial ACT training workshop (days 1–4) occurred while the manualised intervention was being developed. Details of the training workshop are listed in Table 16 (materials are available on request from the corresponding author). Each session comprised a combination of didactic learning through teaching and demonstrations, experiential learning through the personal experience of ACT metaphors and exercises, and practical learning through role plays with other therapists (observed by RG and MS). Following the development of the intervention, therapists attended a 1-day experientially based training workshop on the specific application of ACT to older people with TR-GAD (day 5).

Supervision

Following initial training, therapists were asked to practise delivering ACT to any two service users on their caseload. Fortnightly group supervision and consultation on these cases was provided by telephone by members of the research team (MS and JW) and an independent ACT therapist (Chris Graham). Supervisors were band 8 equivalent or higher clinical psychologists or psychotherapists with a minimum of 5 years’ experience in delivering ACT, as well as experience in supervising ACT. Marc A Serfaty and Julie Loebach Wetherell also had experience of supervising therapists to deliver ACT as part of RCTs. Supervisors observed some of each others’ sessions to ensure that a consistent approach was used with therapists.

After completion of the training workshop, fortnightly group supervision and consultation by telephone was provided by Marc A Serfaty, Julie Loebach Wetherell and Chris Graham for the remainder of the feasibility study. Anonymised supervision notes were taken by the research assistant in each session and made available to therapists and supervisors. The mean number of sessions attended by therapists was 19.1 (SD 9.8), ranging from three sessions (for one therapist who joined the study at a late stage) to 37 sessions.

Therapists

Seventeen therapists received training in ACT, 12 of whom actually delivered the intervention to participants in the study (Table 17). Five therapists dropped out of the study, four without seeing any participants in the study: three because of leaving their job to take up a position elsewhere, one because of retiring and one because of ill health. Therapists were band 7 or 8 clinical psychologists, accredited CBT therapists or counselling psychologists with a minimum of 1 year’s experience in delivering psychotherapy interventions. The majority were based in secondary care in CMHTs for older people (n = 14, 82%), with only three therapists (18%) based in primary care in IAPT services.

Treatment fidelity

All therapy sessions were recorded using encrypted digital voice recorders to monitor adherence to the treatment manual. Ten per cent of sessions were randomly selected and assessed for treatment fidelity by one of two independent ACT therapists with training and experience in delivering ACT using an adapted form of the ACT Treatment Integrity Coding Manual. 99 This is a well-established tool for assessing treatment fidelity in ACT interventions that assesses fidelity in four areas: (1) ACT (six items: deliteralisation/defusion, willingness/acceptance, creative hopelessness/workability, values and goals, committed action, and ACT therapeutic relationship); (2) other (one item: general assessment); (3) anti-ACT (five items: challenging cognitions, using experiential avoidant change strategies, providing a cognitive therapy rationale, implying that thoughts and feelings cause action, and using in-session exposure as a means of decreasing anxiety); and (4) global ratings of adherence (two items: overall adherence to the FACTOID intervention and overall competence of the therapist). All items are rated on a scale of 1 (not at all) to 5 (extensively) with respect to frequency and/or depth of coverage in the session. The mean scores were calculated for each item.

Assessment of treatment fidelity occurred on an ongoing basis throughout the delivery of the intervention. Independent ACT raters were blind to the identity of therapists, and vice versa. Therapists received written feedback from the independent ACT raters as soon as treatment fidelity had been assessed, as well as periodic feedback from Rebecca L Gould with respect to common issues that ACT raters were observing. The random selection of sessions was stratified according to therapist, phase of the intervention (early, middle or late) and phase of study recruitment (early, middle or late), as recommended in psychosocial interventions. 100

In addition to the ACT Treatment Integrity Coding Manual, therapists completed a checklist of ACT components, ACT techniques and themes discussed in each session, together with any ACT-inconsistent deviations from the manual, to facilitate monitoring of treatment adherence.

Usual care

In addition to receiving ACT, all participants received usual care, which was monitored using a modified version of the Client Service Receipt Inventory (CSRI) adapted for use in older people. 87,101 It was anticipated that usual care would comprise GP care with or without multidisciplinary team interventions, including assessment, medication review and management, psychotherapy, occupational therapy and case management. As noted in the eligibility criteria, participants were required to refrain from engaging in other forms of psychotherapy during the receipt of ACT.

Data collection

Data collection was conducted face to face at screening and 0 weeks (baseline) and face to face or by telephone at 20 weeks (follow-up). Data were collected in the participant’s home or at the clinic, depending on patient preference. The following sociodemographic and clinical data were recorded at screening or baseline:

• age, sex, ethnicity, marital status, years of education, highest level of educational qualification, current occupation and highest level of occupational attainment

• psychiatric diagnoses using the MINI97 and the Structured Clinical Interview for DSM-IV Axis II Disorders65

• previous and current treatment for GAD, ongoing medication use including prescribed or illicit substances (dose and frequency) and length of current episode

• GAD using the GAD-782

• global cognition using the SMMSE66

• suicidal ideation, intent and plans

• physical illness and disability using the Cumulative Illness Rating Scale for Geriatrics. 102

Baseline assessments were completed within 2 weeks of starting therapy to account for the fact that all sites refused to fast track study participants to receive therapy because they felt that this would give study participants an unfair advantage over other patients on their therapy waiting lists.

Outcome measures

Measures of bias

Because a drug stabilisation period was not included in the feasibility study, the name, dose and frequency of all psychotropic medication prescribed, and any changes to this, were recorded during the course of the study. This information was extracted from GP medical records, with participants’ consent. This was then included as a covariate in statistical analyses, where necessary. Participants were asked to refrain from engaging in other forms of psychotherapy during the delivery of the intervention because engaging in two types of psychotherapy concurrently may lead to conflicts in therapeutic approaches and goals. Other psychological or psychosocial interventions that participants engaged in during the course of the study were recorded, along with any interventions that participants were referred for after receiving the intervention. Evidence of any adverse effects from the intervention was also monitored and recorded throughout the feasibility study.

Data analyses

Sample size calculation

A sample size of 40 was chosen for a number of reasons. A sample size of 35 is generally recommended for feasibility studies to provide sufficient data and precision of means and variances. 117 Forty participants allows for a 12.5% loss to follow-up based on a preliminary study of ACT in older people with GAD. 27 With respect to session attendance, making the conservative assumption that 80% of participants in the feasibility study will attend ≥ 60% of sessions (i.e. ≥ 10 sessions), a sample size of 35 (after loss to follow-up) would give a 95% CI of 0.63 to 0.92, indicating that we could be 95% certain that at least 63% of the target population would attend ≥ 60% of sessions. Although 100% attendance at ACT sessions was reported in 100% of older people with GAD27 and 78% of older people with depression,59 a conservative estimate was chosen, given the target population (older people with TR-GAD). With respect to recruitment, making a conservative estimate that 60 participants would be eligible to participate in the feasibility study but only 40 would consent to participate, a sample size of 40 would give a 95% CI of 0.53 to 0.78.

Ethics

Ethics approval was granted by the London–Camberwell St Giles Research Ethics Committee on 29 August 2017, with Health Research Authority approval being given on 4 September 2017 (IRAS ID 224111; REC reference 17/LO/1314; see Report Supplementary Material 5).

Participant recruitment and flow

Overall participant recruitment and flow

The flow of participants is shown in Figure 3, and the recruitment rate per month is shown in Appendix 5. The study was due to open for recruitment in November 2017, but this was delayed by 2 months because of delays in NHS recruiting sites providing their confirmation of capacity and capability. A total of 81 potential participants were referred from 2 January 2018 to 31 October 2018, and the 20-week follow-ups occurred between 8 June 2018 and 24 April 2019. On average, 8.1 referrals per month were received. Among the 53 potential participants assessed for eligibility, 11 (21%) were not eligible after pre-screening with the GAD-7 and five others (9%) were excluded for other reasons. A total of 37 of the 42 (88%) eligible participants consented to participate in the study, seven (19%) of whom were lost to follow-up and two (5%) of whom withdrew from the intervention alone. The most common reasons for loss to follow-up or withdrawing from the intervention alone were dissatisfaction with the intervention (n = 4, 44%) and deterioration in physical health (n = 2, 22%). Among the participants who were dissatisfied with the intervention, three stated a preference for a more counselling-type approach in which they could ‘just talk’ and one thought that ACT was not a good fit for them. Among the participants who were dissatisfied with the intervention, two withdrew from both the intervention and the study and two withdrew from the intervention alone (i.e. they completed follow-up outcome measures). Twenty-two participants (59%) attended all 16 sessions, with 27 (73%) attending at least eight sessions.

FIGURE 3.

The CONSORT flow diagram: summary of recruitment and follow-up of participants in the study. This figure has been reported in Gould et al. 94 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. The figure below includes minor additions and formatting changes to the original figure.

This figure has been reproduced from Gould et al. 94 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. The figure includes minor additions and formatting changes to the original.

Session attendance

Overall session attendance

A total of 438 hours of therapy sessions were attended (Figure 4). Twenty-two participants (59%) attended all 16 sessions, with 26 (70%) attending ≥ 10 sessions.

FIGURE 4.

Number of sessions attended by participants.

Waiting time for therapy

The mean waiting time for therapy was 10.2 weeks (SD 8.8 weeks), though there was some variation across sites, with waiting times being the longest for South London and Maudsley NHS Foundation Trust and Camden and Islington NHS Foundation Trust (Table 21). Waiting times were slightly longer for those who were lost to follow-up or withdrew from the intervention.

Demographic and clinical characteristics

Data collection

Participants initially completed a screening assessment in which key demographic and clinical data were collected to determine eligibility for inclusion in the study. Outcome assessments were conducted at 0 weeks (within 2 weeks prior to starting therapy) and 20 weeks. Two participants were withdrawn from the study before completing baseline outcome measures and one participant refused to complete these measures at baseline because of stress. In addition, seven participants were lost to follow-up at 20 weeks. Hence, data collection occurred for 92% (n = 34) of participants at 0 weeks and 81% (n = 30) of participants at 20 weeks. As noted in Waiting time for therapy and Table 21, the mean waiting time for therapy was 10.2 weeks (SD 8.8 weeks), though there was variability across sites. Consequently, the mean number of therapy sessions completed by the 20-week follow-up was 12.1 sessions (SD 3.6 sessions) out of a possible 16 sessions (Table 28).

Examining completion rates further in participants who completed the outcome assessment at 0 and 20 weeks (n = 34 and n = 30, respectively), excellent completion rates were obtained for all questionnaires with the exception of the Credibility/Expectancy Questionnaire (Table 29). The main reason that rates were lower for the Credibility/Expectancy Questionnaire was that some participants reported struggling with estimating responses to the questions and/or did not like having to ‘guess’ responses to the questions.

Primary outcomes

There were four co-primary outcomes: two focusing on feasibility and two focusing on acceptability. Excellent feasibility was demonstrated by a recruitment rate of 93% and a retention rate of 81%, both of which exceeded the targets (Table 30). A high level of acceptability was found with respect to session attendance, as demonstrated by 70% of participants attending ≥ 10 sessions. Rates of satisfaction with therapy were adequate, with 60% of participants scoring ≥ 21 out of 30 on the Satisfaction with Therapy subscale of the Satisfaction with Therapy and Therapist Scale-Revised. However, it should be noted that the median number of sessions completed at 20 weeks was 13.0 sessions (IQR 4.5–14.0 sessions), and 80% (24/30) participants were still undergoing the intervention at the time of the 20-week follow-up assessment. Rates of satisfaction with therapy may have been higher had participants finished receiving their sessions by the time of the 20-week follow-up assessment.

Secondary outcomes

Reliable Change Index

Three patient-reported outcome measures identified from linear mixed-model analyses (Geriatric Anxiety Inventory, Geriatric Depression Scale-15 and Action and Acceptance Questionnaire-II) were analysed using the RCI and CSC analyses. Data used to calculate RCI and CSC scores for each of these measures are outlined in Table 36.

TABLE 36 - Established norms for clinical and non-clinical populations (mean and SD), estimates of reliability (Cronbach’s alpha) and cut-off scores (indicating ‘recovery’) for outcome measures identified in linear mixed-model analyses

N/A, not applicable.

Has to score < 11 points to meet CSC criterion because the established cut-off point is 10/11, where 10 is in the non-clinical range and 11 is in the clinical range. 107

Has to score < 5 points to meet CSC criterion because the established cut-off point is 4/5, where 4 is in the non-clinical range and 5 is in the clinical range. 121

Has to score < 24 points to meet CSC criterion because scores of 24–28 are suggestive of probable clinical distress based on standard cut-off scores on the Beck Depression Inventory-II, General Health Questionnaire-12 and Global Severity Index of the Symptom Checklist-90-Revised. 112

Outcome measure	Geriatric Anxiety Inventory	Geriatric Depression Scale-15	Action and Acceptance Questionnaire-II
Source	Johnco et al.119	Snyder et al.120	Bond et al.112
Population	Older people aged ≥ 60 years with and without comorbid anxiety and mood disorder	Older people aged ≥ 60 years with GAD with and without a comorbid depressive disorder	University students, people in the community and people seeking outpatient psychological treatment
Cronbach’s alpha	0.93	0.73	0.84
Clinical norms
Mean	11.08	7.40	28.30
SD	4.86	3.20	9.90
Non-clinical norms
Mean	0.58	N/A	18.51
SD	1.32	N/A	7.05
Cut-off score (points) used for determining CSC	11a	5b	24c

Among the 29 participants with scores at 0 and 20 weeks, 13 (45%) showed a reliable improvement in scores on the Geriatric Anxiety Inventory at 20 weeks’ follow-up (Figure 5), with six (21%) of these also meeting the criterion for CSC (i.e. scored below the clinical cut-off point). Just over half of participants (n = 15, 52%) showed no reliable change, and only one participant (3%) showed a reliable deterioration in their scores, but this was not clinically significant.

FIGURE 5.

Symptoms of anxiety at 0 and 20 weeks as measured by the Geriatric Anxiety Inventory (n = 29). Participants with the same scores at 0 and 20 weeks share the same marker on the graph. Orange lines indicate the RCI. Further details about the cut-off score for CSC are provided in Table 36.

For the Geriatric Depression Scale-15, just under one-quarter of participants (n = 7, 24%) showed a reliable improvement in scores at the 20-week follow-up, with five (17%) of these also meeting the criterion for CSC (i.e. scored below the clinical cut-off point; Figure 6). Three-quarters of participants (n = 22, 76%) showed no reliable change, and no participants showed a reliable deterioration in their scores.

FIGURE 6.

Symptoms of depression at 0 and 20 weeks as measured by the Geriatric Depression Scale-15 (n = 29). Participants with the same scores at 0 and 20 weeks share the same marker on the graph. Orange lines indicate the RCI. Further details about the cut-off score for CSC are provided in Table 36.

As illustrated in Figure 7, just under one-quarter of participants (n = 7, 24%) showed a reliable improvement in scores on the Action and Acceptance Questionnaire-II at the 20-week follow-up, with four (14%) of these also meeting the criterion for CSC (i.e. scored below the clinical cut-off point). Two-thirds of participants (n = 19, 66%) showed no reliable change, and three participants (n = 10%) showed a reliable deterioration in their scores, but none of these met the criterion for clinically significant change.

FIGURE 7.

Psychological flexibility at 0 and 20 weeks as measured by the Action and Acceptance Questionnaire-II (n = 29). Participants with the same scores at 0 and 20 weeks share the same marker on the graph. Orange lines indicate the RCI. Further details about the cut-off score for CSC are provided in Table 36.

Summary of the main findings

The main purpose of this uncontrolled feasibility study was to assess the acceptability and feasibility of our newly developed ACT intervention for older people with TR-GAD within the NHS. In summary, a high level of feasibility was demonstrated by a recruitment rate of 93% and a retention rate of 81%, both of which exceeded the targets (≥ 80% and ≥ 60%, respectively). A high level of acceptability was found with respect to session attendance, as demonstrated by 70% of participants attending ≥ 60% sessions (i.e. ≥ 10 sessions). Rates of satisfaction with therapy were adequate, with 60% of participants scoring ≥ 21 out of 30 on the Satisfaction with Therapy subscale of the Satisfaction with Therapy and Therapist Scale-Revised, although 80% of participants had not finished receiving their therapy sessions at the time of rating them. Secondary outcome measures further supported the feasibility and acceptability of the ACT intervention for older people with TR-GAD.

Even though the study was not powered to examine clinical effectiveness, there was indicative evidence of improvements in scores on the Geriatric Anxiety Inventory (–2.30 points, 95% CI –3.83 to –0.76 points), Geriatric Depression Scale-15 (–2.04 points, 95% CI –3.31 to –0.77 points) and Acceptance and Action Questionnaire-II (–3.93 points, 95% CI –7.16 to –0.70 points) from the 0 to the 20-week follow-up. Reliable improvements in scores were found in 45% of participants on the Geriatric Anxiety Inventory and in 24% of participants on both the Geriatric Depression Scale-15 and the Acceptance and Action Questionnaire-II. These results were particularly impressive given that all participants had failed to respond to prior pharmacological and/or psychological therapy for GAD. Furthermore, there was indicative evidence of improvement in scores on the Geriatric Depression Scale-15, even though participants were selected on the basis of GAD and not depressive symptoms.

These improvements are partially consistent with the results of a very small study of ACT for older people with GAD in which participants showed significant pre–post improvements in worry and depressive symptoms but not in anxiety. 27 The small discrepancy in findings might be explained by a number of issues in the previous study: (1) the Hamilton Anxiety Rating Scale was used to assess anxiety rather than a measure designed specifically for older people, such as the Geriatric Anxiety Inventory, (2) older people were not specifically included on the basis of TR-GAD and (3) the sample size was very small (n = 7). The improvements reported here are also partially consistent with the results of a small RCT of ACT compared with a waiting list control for older people with symptoms of depression and/or anxiety who were living in long-term care facilities. 60 This study reported significant between-group differences in depression but not anxiety, which the authors suggested was due to mean scores for anxiety being below clinical cut-offs at baseline.

A number of key issues should be highlighted when considering these results. First, this small, uncontrolled feasibility study cannot answer questions concerning the efficacy or clinical effectiveness of the ACT intervention. Nevertheless, these results suggest that the intervention may have the potential to reduce anxiety and depression and to increase psychological flexibility in older people with TR-GAD.

Second, symptomatic reductions in anxiety and depression were observed, even though this is not the main focus of ACT. The focus in ACT is on ‘living better’ rather than ‘feeling better’, and any symptomatic reductions are generally seen as a by-product of therapy. A measure that tapped into ‘living better’ may have demonstrated even larger improvements in scores.

Third, evidence of improvements in scores were found on the Acceptance and Action Questionnaire-II. This is a measure of psychological flexibility that primarily examines the degree to which a person is engaging in experiential avoidance. It does not measure other ACT processes and, therefore, it does not provide a comprehensive assessment of change in ACT processes. It may be that greater changes in ACT processes would have been observed with a broader measure such as the Comprehensive Assessment of Acceptance and Commitment Therapy Processes (CompACT) questionnaire. 122 This is a 23-item measure that examines (1) willingness to experience thoughts, feelings, sensations, etc., without trying to control, avoid or get rid of them; (2) mindful attention to current behaviour or actions; and (3) engagement in meaningful activity or valued actions.

Fourth, rates of satisfaction with therapy were found to be adequate on the Satisfaction with Therapy subscale of the Satisfaction with Therapy and Therapist Scale-Revised, with 60% of participants scoring ≥ 21 out of 30 points. As previously noted in Outcome measures, Primary outcome measures, there is no set definition of what constitutes ‘satisfactory’ and so this was arbitrarily defined as a total score of ≥ 21 out of 30 points on this subscale. If, however, a total score of ≥ 18 out of 30 points had been chosen (which corresponds to a ‘neutral’ rating on all items, i.e. the lower limit of acceptability), then 70% of participants would have rated the therapy as ‘satisfactory’. There are a number of reasons why rates of satisfaction with therapy may have been lower than expected. It may be that the intervention requires further refinement to make it acceptable to a larger proportion of older people with TR-GAD. This is explored in Chapter 6. It may also have been due to the fact that 80% of participants had not finished receiving their therapy sessions by the 20-week follow-up assessment. Finally, it may have been a result of the type of questionnaire chosen to assess satisfaction with therapy. The Satisfaction with Therapy and Therapist Questionnaire-Revised was developed for ‘fix it’ therapies (i.e. those that are focused on getting rid of problematic thoughts, feelings and situations) and, hence, it is less suitable for therapies that are focused on ‘living better’ rather than ‘feeling better’. Higher rates of satisfaction with therapy may have been found with a questionnaire focused on changes in life satisfaction rather than symptomatic change.

Finally, this study demonstrated not only the acceptability and feasibility of the intervention and study but also that it is possible to train novice ACT therapists to deliver this intervention to older people with TR-GAD within the NHS. This is consistent with previous findings. 27,123 No other studies have examined pharmacological or psychological treatments for older people with TR-GAD.

Research implications

This study examined whether or not it was feasible to deliver an ACT-based intervention to older people with TR-GAD within the NHS, and it explored the acceptability of this intervention. Although it was not powered to demonstrate clinical effectiveness, results indicated that reliable changes in key patient-reported outcome measures were observed in some participants. Most notably, reliable improvement in scores on the Geriatric Anxiety Inventory was observed in 45% of participants at the 20-week follow-up. This suggests that there is potential for clinical effectiveness in this population, which should be tested in a future substantive trial.

A wide variation in the reported number of years experiencing difficulties with worrying was observed in the current study: just over one-third of participants (n = 13, 35%) reported experiencing difficulties for the past 1–5 years, and just over one-third (n = 13, 35%) reported experiencing difficulties for > 30 years. A bimodal distribution of age at onset has been found in other studies of GAD in working age adults and older people. 124–126 This suggests that there may be two different subpopulations here: (1) a subpopulation in which GAD is lifelong and (2) a subpopulation in which GAD is more recent, because excessive worry and anxiety are required to be present only for at least 6 months to meet International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10),127 and DSM-V,128 diagnostic criteria (along with the presence of other symptoms). Alternatively, it may be that some people in the latter subpopulation also experience lifelong GAD but are able to use coping strategies to ‘mask’ GAD symptoms during their working life (e.g. overengagement in activity or work as a means of distracting themselves from worries). When these coping strategies are no longer available owing to age-related factors (such as retirement or physical health problems that do not permit overengagement in activity), then GAD symptoms are ‘unmasked’, giving the appearance of a recent onset. Whether or not there are qualitative differences between these subpopulations with respect to demographic, clinical and personality characteristics, as well as treatment response, should be explored further in a larger study. Furthermore, subgroup analyses that account for the age at onset of GAD should be included in any future RCT examining the clinical effectiveness of ACT for TR-GAD, given that higher rates of psychiatric comorbidity, psychotropic medication use and severe worry have been reported in older people with an earlier rather than a later onset (i.e. before vs. after the age of 50 years). 124

Following on from this, it has been argued that ‘the letters GAD would better stand for godawful diagnosis’129 and that this diagnostic category should be abandoned. Three possible diagnostic replacements have been suggested for GAD: (1) a diagnosis of mixed anxiety–depression due to the fact that these symptoms commonly occur together, so-called cothymia;130 (2) a diagnosis of ‘general neurotic syndrome’ or ‘general nervous syndrome’ due to the overlap between personality status and GAD; and (3) an adjustment disorder due to experiencing a major life event (i.e. life event-precipitated GAD).

Data reported here suggest that none of these possible diagnostic replacements applies to all of the participants in the current study, and many participants would fit more than one category. Sixty-two per cent (n = 23) of participants met MINI diagnostic criteria for GAD plus major depressive episode with or without melancholic features or mood disorder with psychotic features. Only a small proportion of participants (n = 6, 16%) met screening criteria on the Structured Clinical Interview for DSM-IV Axis II Disorders, with only two people reporting a comorbid diagnosis of personality disorder. However, a larger proportion of participants (n = 12, 32%) were unable to report a period of time in their life during which they experienced no difficulties with excessive worrying (suggestive of a recurrent, life-long condition that Tyrer129 would describe as ‘general neurotic syndrome’ or ‘general nervous syndrome’). Just over one-third of participants (n = 13, 35%) reported that a major or stressful life event had preceded excessive worrying and anxiety. It is possible that the response to pharmacological and/or psychological treatment and, hence, management of excessive worry and anxiety, may differ across these possible subpopulations. However, the fact that 43% (n = 16) of participants fit more than one category may blur any such distinctions. Thus, again, an examination of demographic, clinical and personality characteristics, as well as treatment response, should be explored further in a larger study.

Strengths of the study

To our knowledge, this is the first study to report the acceptability and feasibility of a psychological intervention that was developed specifically for older people with TR-GAD for delivery within the NHS. Other strengths of the study include the following:

• We were able to demonstrate that it is feasible to successfully recruit participants whom many clinicians find challenging to work with owing to chronicity and resistance to treatment of GAD symptoms.

• Engagement was high and drop-out was relatively low given that participants were experiencing difficult to treat symptoms by definition.

• We demonstrated strong evidence of feasibility and acceptability for the majority of primary and secondary outcomes.

• Given the chronicity and resistance to treatment of GAD symptoms, the fact that reliable improvements in scores on the Geriatric Anxiety Inventory were found in almost half of older people with TR-GAD was impressive.

• We were able to train therapists who are new to the ACT model to deliver the intervention within the NHS, with excellent levels of adherence and competence.

• We established a therapist supervision model that helped therapists to feel supported in the delivery of the intervention, which could be easily adopted in a future substantive trial.

Limitations of the study

There are a number of limitations of the current study. First, there was no control condition; therefore, it is uncertain whether or not any changes in outcome measures were related to the intervention or were a product of other non-specific therapeutic factors such as social support or spontaneous recovery. Three participants reported withdrawing from the intervention as they would have preferred a more counselling-type of approach in which they could ‘just talk’. This suggests that an appropriate comparison could have been a talking control condition in which participants were allowed to ‘just talk’. It has previously been shown that it is not only feasible to deliver such a talking control condition to older people with depression but also acceptable to them. 131 There would be a number of advantages of including a talking control condition in a future RCT of ACT for older people with TR-GAD. This would not only control for non-specific therapeutic factors such as social support but could also reduce attrition due to resentful demoralisation, which can occur when participants are allocated to a non-active control condition such as treatment as usual. An additional benefit is that it may be easier to recruit participants with TR-GAD into a RCT of ACT if it is compared with a talking control condition. The very nature of GAD means that potential participants may worry about which arm they would be randomly allocated to in a RCT. Consequently, they may decline to participate because they want to avoid this uncertainty. Including a talking control condition rather than a non-active control such as treatment as usual may remove some of this uncertainty.

Another limitation is that the majority of participants were white/white British women who would be categorised as ‘younger old’ (i.e. in their 60s and 70s). Although just over half of participants were in their 70s, we recruited as many participants in their 80s as those in their 60s, which is acceptable for late-life mental health studies. Furthermore, although the number of participants from ethnic minorities was small and clearly under-representative, we demonstrated that we were able to recruit from these populations. In addition, although the majority of participants were women, this is unsurprising given that being female is one of the known risk factors for anxiety disorders in older people,90 and the older that people are, the more likely they are to be female. However, these sampling issues do mean that the results cannot be generalised to a broader population, particularly ‘older old’ (i.e. in their 80s or older) men from ethnic minorities. Recruiting ‘harder to reach’ older people is a common problem in research trials in general, and identifying successful strategies for recruiting from these populations would be important in any future study. Such strategies may include those recommended by the TIBaR model for recruiting ‘hard-to-reach’ older people: (1) build up Trust, (2) offer Incentives, (3) identify individual Barriers and (4) be Responsive. 132 They may also include strategies identified from RCTs of different recruitment approaches. For example, a previous study of physical activity in older people found that an invitation letter plus a patient information sheet that was followed up by up to four attempts at telephone contact was the most effective recruitment strategy. 133

A further limitation is the fact that participants were recruited only from urban and suburban areas in the London region, because this was a feasibility study, thus limiting external validity. Clearly, recruiting participants from multiple sites across the UK, including those in rural regions, would enhance the representativeness of findings in a future substantive trial.

The fact that outcome measures were assessed at only 0 and 20 weeks is an additional limitation of the current study. Some participants had not finished their ACT sessions by 20 weeks owing to the long waiting list for therapy at some sites, which may have negatively affected change scores. Furthermore, no follow-up after 20 weeks was conducted and, therefore, it is uncertain whether or not any gains were maintained beyond 20 weeks. It is also not clear whether or not a follow-up after 20 weeks would be feasible in a future substantive trial. Clearly, including an additional follow-up assessment in a future substantive trial would help to clarify these uncertainties.

Another limitation is that the sample size was relatively small, although this was consistent with previous recommendations of 24–35 participants for pilot and feasibility studies to provide sufficient data and precision of means and variances. 117,134,135 Consequently, conclusions should be drawn with caution, particularly those relating to patient-reported outcomes.

A final limitation relates to the assessment of treatment fidelity in the delivery of training and intervention sessions. We did not conduct any direct visual observation of intervention sessions to assess treatment fidelity for pragmatic reasons (including the desire to avoid anxiety-provoking study procedures in an already anxious group and to reduce potential demand characteristics). 136 Instead, an independent ACT therapist assessed treatment fidelity through audio-recordings of sessions – a standard approach that has been used in previous RCTs. 87 Although the training sessions were developed by members of the research team with knowledge and experience in delivering ACT and training therapists to deliver ACT, we did not directly assess the fidelity of the training sessions. Therefore, the degree of adherence to the ACT model could not be ascertained.

Outcome measures

All reporting is in accordance with the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist, where possible, because this was neither a randomised trial nor a full economic evaluation and did not include a cost-effectiveness analysis (see Appendix 6).

Data analysis

Missing data

Data were collected from 34 participants at 0 weeks and 30 participants at 20 weeks. As shown in Table 37, there were low rates of missing data for the EQ-5D-5L and the modified CSRI. Some participants reported finding it difficult to estimate their health using a visual analogue scale in the EQ-5D-5L.

Intervention costs

A detailed breakdown of the costs associated with the delivery of the intervention, including training and supervision of therapists, is presented in Table 38. The highest costs were for training (£31,802.34 without overheads and £53,235.50 with overheads) and the lowest costs were for supervision (£15,403.22 without overheads and £25,062.00 with overheads). With overheads, the estimated total cost was £112,866.75, equating to a mean cost of £3050.45 per participant. Without overheads, the estimated total cost was £68,200.41, equating to a mean cost of £1843.25 per participant.

Quality of life

Health and social care resource use

Low rates of service use were reported at both time points (Table 42), with very little service use being self-reported at 20 weeks. The most common types of services used were outpatient services (n = 11, 33%), with psychiatric outpatient services being the most frequently reported (n = 8, 24%). Community mental health centre was the only category endorsed in day activity services. Community care services mostly featured those focused on mental health [e.g. Care Programme Approach (CPA) key worker, community psychiatric nurse]. No use of criminal justice services was reported at 0 or 20 weeks.

The total cost of service use per participant is reported in Table 43, broken down into community care, secondary care and medication costs. Costs appear to be lower at the 20-week follow-up than at baseline. The higher costs at baseline are driven by a couple of individuals with very high costs across all types of services, including multiple psychiatric outpatient appointments and many contacts with a CMHT and CPA key worker, as well as the prescription of various medications. Without a control group and given the small sample size and skewed nature of the costs, the reduction in costs between 0 and 20 weeks cannot be conclusively attributed to the intervention.

Main findings and associated factors that would need to be considered in a future substantive trial

Excellent completion rates were obtained for the EQ-5D-5L and the modified CSRI, indicating their feasibility for use in a future substantive trial. In support of this, these instruments have been successfully applied in previous studies of cost-effectiveness in older people. 87,147 Furthermore, a systematic review of instruments for measuring quality of life outcomes for economic evaluations recommended the EQ-5D-5L for use with older populations. 148

Strengths and limitations

No evidence currently exists with respect to the cost-effectiveness of psychological or pharmacological interventions for older people with TR-GAD. This study represents the first steps to addressing this. However, there are a number of limitations of the study, as outlined in Chapter 4 and in this section.

First, we were unable to perform a cost-effectiveness analysis because of the lack of a control group. Consequently, this would need to be performed alongside a future substantive RCT.

Second, although health-related quality of life was measured using the EQ-5D-5L in the current study, no other measure of quality of life was included. Ideally, a measure that focused more on psychological health as opposed to physical health should have been included, in addition to the EQ-5D-5L, because it would not have been overly influenced by physical health problems that are common in this population. Such an example is the McGill Quality of Life Questionnaire-Revised,154 which is a global measure of quality of life that has satisfactory psychometric properties. This questionnaire places relatively less emphasis on physical well-being than other measures of quality of life, which means that scores would not be overly influenced by physical health problems. Furthermore, this measure includes items that are consistent with the treatment goals of ACT: assessing participants’ ability to undertake personally meaningful and purposeful activities and to achieve life goals in the presence of TR-GAD. However, this questionnaire cannot be used to calculate QALYs and so could not contribute to a cost–utility analysis alongside the EQ-5D-5L.

Other quality of life measures that could be used in addition to the EQ-5D-5L (or an equivalent instrument) to assess the psychological component of quality of life and contribute to a cost–utility analysis include the Short Form questionnaire-36 items (SF-36) and the ICEpop CAPability measure for Older people (ICECAP-O). 155,156 The SF-36 can be used to calculate QALYs whereas the ICECAP-O can be used to calculate capability-adjusted life-years and aims to capture well-being broader than just health. The ICECAP-O would be preferred over the SF-36 because the mental health component of the SF-36 focuses more on symptoms of psychological distress (e.g. feeling nervous or downhearted and ‘blue’) than the ICECAP-O. This may be more appropriate for ‘fix it’ therapeutic approaches that are aimed at helping people to feel better (e.g. CBT) rather than therapeutic approaches that are aimed at helping people to live better alongside their difficulties (e.g. ACT). By contrast, the ICECAP-O focuses on love and friendship, the ability to think about the future without concern, the ability to do things that make a person feel valued, enjoyment and pleasure, and independence. Some of these items fit better with the aims of ACT and, therefore, may be more sensitive to change than the SF-36.

Finally, one of the limitations noted in Chapter 4 was the fact that the majority of participants were in their 60s or 70s, which meant that the results cannot be generalised to those in their 80s or older. The small number of participants reporting at least moderate problems with self-care on the EQ-5D-5L at 0 and 20 weeks (n = 4) further suggests that results may not be representative of frailer older people living in the community.

Design

The study used an explorative, flexible, qualitative research design to gather in-depth data. 157

Participants and recruitment procedures

Older people with TR-GAD who had participated in the uncontrolled feasibility study and therapists who had delivered ACT in the study were invited to participate in individual qualitative interviews. Purposive sampling was conducted on the basis of sex, ethnicity, recruitment source and session attendance to explore a range of perspectives. Participants who withdrew from the intervention or study were invited to participate in qualitative interviews; all declined. All therapists were invited to participate in individual qualitative interviews to examine how the intervention was delivered in practice.

Procedure

Semistructured interviews were conducted with 18 older people with TR-GAD and 11 therapists until data saturation was reached. The topic guide for older people with TR-GAD was used flexibly to explore participants’ experiences of receiving ACT, how helpful they found it, how easy it was to understand, the degree to which it met their needs, any difficulties experienced, any facilitators of and barriers to engaging in ACT, and any recommendations for revising the intervention (see Report Supplementary Material 6). The topic guide also explored any changes that people had experienced in their lives as a result of receiving ACT, as well as opinions in relation to the design of a future substantive trial.

The topic guide for therapists was used flexibly to examine therapists’ experiences of delivering ACT, how easy it was to deliver ACT to older people with TR-GAD, how understandable and how suitable it was for older people with TR-GAD, any difficulties experienced, any facilitators of and barriers to engaging in ACT, and any recommendations for revising the intervention. The topic guide also explored any changes that they had noticed in participants as a result of receiving ACT, as well as in their own professional practice, and opinions about the design of a future substantive trial.

Interviews with older people with TR-GAD were completed after the participant had finished receiving their therapy sessions. Interviews with therapists were completed once they had finished delivering all of their therapy sessions to study participants. All interviews were conducted by one researcher (KK) and were audio-recorded and transcribed verbatim. All participants consented to participate in semistructured interviews.

Data analyses

Thematic analysis was used to analyse qualitative data. This method of analysis was chosen as it allows accessibility and flexibility in its approach. 158 Thematic analysis can be used to identify and report patterns in the data and to interpret various aspects of a research topic. 159 One researcher (KK) listened to all audio-recordings and repeatedly read the transcripts and research notes to familiarise herself with the data. Key issues, recurrent themes and interpretations were noted and discussed in supervision.

Ethics

Ethics approval was granted by the London–Camberwell St Giles Research Ethics Committee on 29 August 2017 and Health Research Authority approval granted on 4 September 2017 (IRAS ID 224111, REC reference 17/LO/1314; see Report Supplementary Material 5).

Older people with treatment-resistant generalised anxiety disorder