How should psychological interventions for distressing voices be delivered? A comparison of outcomes for patients who received interventions remotely or face-to-face within routine clinical practice

ABSTRACT

Aims

This service evaluation aimed to compare the outcomes for interventions for distressing voices that were delivered face-to-face (F2F) or remotely. It was hypothesised that F2F interventions would generate greater reductions in the primary outcome of the negative impact of voices, relative to remote delivery.

Methods

Sixty-three patients (33 F2F; 30 remote) completed a baseline assessment, 4–8 sessions of intervention and a post-intervention assessment. At both assessments, patients completed questionnaires that measured voice hearing experiences, recovery, depression and anxiety. The primary outcome was negative impact of voices measured by the Hamilton Programme for Schizophrenia Voices Questionnaire-Emotional Subscale (HPSVQ-ES).

Results

A non-significant difference was found between groups. The Minimum Clinically Important Difference for the primary outcome was met for remote delivery and was within the confidence intervals for F2F. Given a larger sample size, these findings suggest that interventions may have generated clinically meaningful benefits, irrespective of the mode of delivery.

Discussion

Given the small sample size, it may be premature to draw any conclusions from this evaluation. However, the promising outcomes across modes of delivery suggest that it may be appropriate to offer patients a choice of how they wish to receive interventions for distressing voices.

KEYWORDS:

• Voice hearing
• CBT
• pandemic
• face-to-face
• remote

The remote delivery of psychological interventions has drawn much attention in recent years (Fairburn & Patel, 2017). During the Covid-19 pandemic, this interest has accelerated (Simpson et al., 2021) as most psychological interventions were delivered remotely (e.g. Pierce et al., 2021). Pre-pandemic data suggested that interventions delivered remotely for common mental health problems were non-inferior in comparison to interventions delivered face-to-face (e.g. health anxiety [Axelsson et al., 2020;] and depression [Luo et al., 2020]). However, limited data are available to inform the debate about the benefits of remotely delivered interventions for patients with severe mental health problems (Kopelovich & Turkington, 2021). Whilst emerging data is suggesting that psychological interventions can be beneficial when delivered digitally (Garety et al., 2021) and remotely (Ellett et al., 2022) to patients with severe mental health problems, data is required to facilitate comparisons with interventions delivered face-to-face. This service evaluation sought to offer comparative data as a psychological interventions service for patients with severe mental health problems moved from face-to-face delivery to remote delivery due to the pandemic.

The Sussex Voices Clinic (SVC) seeks to increase access to evidence-informed interventions for patients distressed by hearing voices (https://www.sussexpartnership.nhs.uk/sussex-voices-clinic). This is achieved by working across diagnoses (Morrice et al., 2022) and supporting a wider workforce of clinicians to deliver interventions (Clarke et al., 2021). During the COVID-19 pandemic, the maintenance of access to interventions necessitated remote delivery via phone or videoconferencing. This service evaluation used data collected during routine clinical practice in a secondary care mental health service to compare outcomes of patients distressed by hearing voices who completed interventions either face-to-face (F2F; prior to the pandemic) or remotely (during the pandemic). Given the existing evidence for effectiveness when interventions are delivered F2F (e.g. Lincoln & Peters, 2019), it was hypothesised that in comparison to remote delivery, F2F delivery would generate greater improvements on the negative impact of voices (the primary hypothesis) and on recovery, anxiety and depression (secondary hypotheses).

Method

Design

This was a service evaluation comparing the pre-post outcomes of patients who completed a cognitive behavioural therapy (CBT)-informed intervention delivered either F2F or remotely within SVC, a specialist trans-diagnostic outpatient service within Sussex Partnership NHS Foundation Trust (SPFT). SVC trains and supervises clinicians within secondary care mental health services to deliver brief and structured interventions to patients within their services who are distressed by hearing voices. NHS Research Ethics Committee approval was not required for the study because it was completed as a service evaluation of routine practice within a clinical service (Department of Health, 2017). Permission to conduct the service evaluation was granted by the Quality Improvement Support Team in SPFT on 4^th May 2021 and by the School of Psychology at the University of Sussex on 16^th November 2021.

Patients

SVC offers interventions to patients aged 18 years or over who report voice hearing experiences that are having a “negative impact” (operationalised at baseline assessment as a score of 8 or more on the Hamilton Program for Schizophrenia Voices Questionnaire (HPSVQ) Emotional Subscale (HPSVQ-ES: Van Lieshout & Goldberg, 2007)). To be included within this evaluation, patients were additionally required to have completed an intervention that was delivered either F2F or remotely and offered a full dataset. A total of 63 participants met the inclusion criteria between March 2018 and July 2021: 33 patients received an intervention delivered F2F between March 2018 and July 2019 (a time-period when interventions were exclusively offered face-to-face); and 30 received an intervention delivered remotely between March 2020 and July 2021 (a time-period when interventions were exclusively offered remotely). Consistent with the emerging literature suggesting that voice hearing experiences are similar across diagnoses (Strawson et al., 2022; Waters & Fernyhough, 2017), diagnosis was not an inclusion criterion for this evaluation.

Materials

Patients completed four measures: the HPSVQ-ES, a 4-item measure of the negative impact of voices (Van Lieshout & Goldberg, 2007); the Choice of Outcome In CBT for Psychoses (CHOICE), an 11-item measure of patient-rated recovery (Webb et al., 2021); the Generalised Anxiety Disorder Assessment (GAD-7), a 7-item measure of anxiety (Spitzer et al., 2006); and the Patient Health Questionnaire (PHQ-9), a 9-item measure of depression (Kroenke et al., 2001). Assessments were completed at baseline and post-intervention by clinic assistants who were not involved with the delivery of interventions.

Interventions

Patients were offered 4–8 sessions of a CBT-informed intervention for distressing voices, either Coping Strategy Enhancement (CSE – a 4-session intervention seeking to enhance the effectiveness of naturally occurring coping strategies, Hayward et al., 2018) or Guided self-help for voices (the GiVE intervention – an 8-session intervention seeking to enhance coping, re-evaluate beliefs about self and voices, and facilitate assertive relating, Hazell et al., 2018). The interventions were delivered by clinicians with varying levels of therapy training, each of whom had attended an SVC training workshop in the delivery of the intervention and was supervised within SVC. Both the CSE and GiVE interventions can generate benefits to patients when delivered by briefly trained therapists (Clarke et al., 2021; Hayward et al., 2021).

Statistical analysis

The primary hypothesis was tested using a two-way mixed analysis-of-covariance (ANCOVA) model with negative impact of voices (HPSVQ-ES) as the dependent variable with intervention (F2F or remote) and time (pre- or post-intervention) as independent variables. Gender was included as a covariate due to the literature suggesting a difference between males and females in HPSVQ-ES scores (Chung et al., 2019; Morokuma et al., 2017). Other patient demographics were tested for inclusion in the ANCOVA as covariates using independent samples t-tests. A statistically significant time*intervention interaction in the model indicated greater change in one group compared to the other. Secondary hypotheses were tested using the same approach (with the addition of a formal test for gender). To address the bias induced by missing data, multiple imputation was performed followed by a sensitivity analysis. All statistical tests were significant at the 5% alpha level and all analyses were carried out using SPSS Version 26.

Results

Patient characteristics

Thirty-two patients (50.8%) were female, 24 (38.1%) were male and 7 (11.1%) preferred not to disclose their gender. Nine were employed (14.3%), 44 (69.8%) were unemployed and 10 (15.9%) did not disclose their employment status. Finally, 46 (73%) patients were single and 14 (22.2%) were in a relationship and 3 (4.8%) did not disclose their relationship status. These figures were broadly similar across both the F2F and remote groups.

Primary analysis

The final two-way mixed ANCOVA model to explore the primary hypothesis for dependent variable HPSVQ-ES included gender and number of sessions as covariates. Relationship status and employment status were also tested as potential covariates but showed non-significant between-group differences in HPSVQ-ES scores. The results (see Tables 1 and 2) showed no difference by intervention group in the pre-post change for the primary outcome. This was indicated by a non-significant time*intervention interaction on HPSVQ-ES scores [b = −1.97, 95% CI [−4.6, 0.7], p = .146]. However, there was a statistically significant gender effect [F(1, 52) = 6.45, p = .014] and mean HPSVQ-ES scores for females were higher than those for males; this finding is in line with the literature (e.g. Hayward et al., 2016).

Table 1. Descriptive summary of outcome measures by intervention group and pre-post intervention differences using original data.

Outcome measure	Group	Count n	Baseline M (SD)	Post-intervention M (SD)	M pre-post difference	95% CI [LL, UL]
HPSVQ emotional subscale score	Face-to-face	28	11.9 (3.6)	10.9 (4.1)	−1	[−1.1, −0.4]
	Remote	28	12.9 (2.5)	10.6 (3.7)	−2.3	[−2.9, −2.3]
GAD-7 score	Face-to-face	26	14.9 (5.6)	13.7 (6.6)	−1.2	[−1.2, −0.4]
	Remote	23	14.2 (6)	12.5 (7)	−1.7	[−2.5, −1.8]
PHQ-9 score	Face-to-face	26	19.7 (5.1)	17.4 (7.1)	−2.3	[−2.2, −0.7]
	Remote	21	19.8 (5.9)	17.1 (8.2)	−2.7	[−4.5, −2.7]
CHOICE mean	Face-to-face	27	3.8 (2)	4.4 (2.3)	0.6	[0.4, 0.6]
	Remote	23	4.4 (1.6)	5.1 (1.8)	0.7	[0.8, 1]
CHOICE goal rating	Face-to-face	29	2.7 (2.1)	4.9 (2.5)	2.2	[2, 2.1]
	Remote	26	2.9 (2)	5.5 (1.9)	2.6	[2.6, 2.8]

n, number of participants; M (SD), mean and standard deviation; M, mean; 95% CI [LL, UL], 95% confidence interval lower level, upper level.

Table 2. Results of time*intervention interactions via two-way mixed ANOVA models based on original data and multiple imputation data.

		Original data				Multiple imputation data
Outcome measure	b		SE	95% CI [LL, UL]	p	b	SE	95% CI [LL, UL]	p
HPSVQ emotional subscale score	−1.97		1.36	[−4.63, 0.69]	.146	−1.83	1.35	[−4.48, 0.82]	.175
GAD-7 score	−2.84		2.6	[−7.94, 2.26]	.275	−1.78	2.5	[−6.68, 3.11]	.476
PHQ-9 score	−1.05		2.81	[−6.55, 4.45]	.709	−1.41	2.61	[−6.52, 3.7]	.588
CHOICE mean	0.52		0.81	[−1.06, 2.1]	.52	0.3	0.79	[−1.25, 1.85]	.701
CHOICE goal rating	0.3		0.89	[−1.44, 2.04]	.736	0.1	0.89	[−1.65, 1.85]	.91

b, coefficient; SE, standard error; 95% CI [LL, UL], 95% confidence interval lower level, upper level; p, level of significance.

Secondary analysis

The secondary hypotheses were tested using two-way mixed ANCOVAs with number of sessions as a covariate (see Tables 1 and 2). The time*intervention interaction effect for HPSVQ-ES scores can be seen in Table 1.

Treatment response rates

The minimal clinically important difference for the HPSVQ-ES is a two-point reduction or increase. In the F2F group, 14 (42%) patients experienced a reduction from baseline at or above the MCID, 8 (24%) experienced an increase from baseline at or above MCID and 11 (33%) did not meet MCID. Similarly, in the remote group, 17 (57%) patients experienced a reduction at or above MCID, 3 (10%) experienced an increase at or above MCID and 10 (33%) did not meet MCID.

Sensitivity analysis

Table 2 displays the time*intervention interactions for each outcome measure using ANCOVA based on original data (complete cases) and multiple imputation findings. The two sets of results were consistent, and no conclusions were altered after addressing the potential missing data bias.

Discussion

This service evaluation explored the outcomes of patients who completed CBT-informed interventions for distressing voices within routine clinical practice. Outcomes were compared for patients who received interventions either F2F (before the COVID-19 pandemic) or remotely (during the COVID-19 pandemic), and it was hypothesised that interventions delivered F2F would generate more favourable outcomes. Primary and secondary hypotheses were not supported, suggesting that the method of delivery did not significantly influence the outcomes. These findings offer some support to the possibility that, like common mental health problems (Luo et al., 2020), the benefits of interventions delivered remotely for severe mental health problems may not be inferior to the benefits of interventions delivered face-to-face.

This evaluation had several limitations that require the findings to be interpreted with caution. Firstly, patients were allocated to receive different interventions based on clinical judgement and interventions were offered by clinicians with varied experience of delivering therapies. Whilst this was appropriate in the context of the current service evaluation, this means that the groups cannot be assumed to be comparable. Future studies would need to allocate randomly. Secondly, the post-intervention assessments could not be conducted blind – increasing the likelihood that the assessments may have been influenced by the desire of the assessors to elicit favourable outcomes. Thirdly, the service evaluation was not controlled, creating the un-evaluated possibility that any benefits occurred naturally over time. Finally, there was no attempt to assess the sustainability of benefits over a follow-up period – leaving the unanswered question about a relationship between method of delivery and the maintenance of gains. These limitations are common to studies that use service evaluation methodology and are balanced to some extent by the ability of this methodology to offer ecologically valid data from the “real world”.

In the context of these limitations, the implication of this evaluation is that people should be offered future choices about how they receive interventions within SVC. If they choose to receive interventions remotely, this may have positive implications for cost-effectiveness (e.g. there may be less demand on the use of mental health centres) and increasing access (e.g. the time that therapists spend travelling can be allocated to the delivery of interventions). However, assumptions should not be made about the choices that patients will make, as voice hearing patients surveyed during the pandemic (N = 335) expressed a clear preference for interventions to be delivered F2F (Berry et al., 2022). The expression of other clear preferences (e.g. the timing [weekday afternoons preferred] and location [clinical settings preferred] of sessions) suggests that many patients have preferences for how they receive psychological interventions, and these preferences should be sought and acted upon, where possible. Future research should robustly explore the experiences and outcomes generated by interventions delivered in different formats to patients distressed by hearing voices.

Disclosure statement

No potential conflict of interest was reported by the authors.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.