Synthesis of triphenylphosphonium dibenzothiophene S-oxide derivatives and their effect on cell cycle as photodeoxygenation-based cytotoxic agents

Article abstract of DOI:10.1016/j.bioorg.2020.104442

Photodeoxygenation of Dibenzothiophene-S-oxide (DBTO) in UV-A light produces atomic oxygen [O(³P)] and the corresponding sulfide, dibenzothiophene (DBT). Recently, DBTO has been derivatized to study the effect of UV-A light-driven photodeoxygenation in lipids, proteins, and nucleic acids. In this study, two DBTO derivatives with triphenylphosphonium groups were synthesized to promote mitochondrial accumulation. The sulfone analogs of these derivatives were also synthesized and used as fluorescent mitochondrial dyes to assess localization in mitochondria of HeLa cells. These derivatives were then used to study the effect of photodeoxygenation on MDA-MB-231 breast cancer cell line using cell viability assays, cell cycle phase determination tests, and RNA-Seq analysis. The DBTO derivatives were found to significantly decrease cell viability only after UV-A irradiation as a result of generating corresponding sulfides that were found to significantly affect gene expression and cell cycle.

Full text of DOI:10.1016/j.bioorg.2020.104442

Articles
Effectiveness of an integrated telehealth service for patients
with depression: a pragmatic randomised controlled trial of
a complex intervention
Chris Salisbury, Alicia O’Cathain, Louisa Edwards, Clare Thomas, Daisy Gaunt, Sandra Hollinghurst, Jon Nicholl, Shirley Large, Lucy Yardley,
Glyn Lewis, Alexis Foster, Katy Garner, Kimberley Horspool, Mei-See Man, Anne Rogers, Catherine Pope, Padraig Dixon, Alan A Montgomery
Summary
Background Many countries are exploring the potential of telehealth interventions to manage the rising number of Lancet Psychiatry 2016;
3: 515–25
people with chronic disorders. However, evidence of the effectiveness of telehealth is ambiguous. Based on an
evidence-based conceptual framework, we developed an integrated telehealth service (the Healthlines Service) for
Published Online
April 27, 2016
chronic disorders and assessed its effectiveness in patients with depression. We aimed to compare the Healthlines
http://dx.doi.org/10.1016/
Depression Service plus usual care with usual care alone.
S2215-0366(16)00083-3
See Comment page 490
Methods This study was a pragmatic, multicentre, randomised controlled trial with participants recruited from
43 general practices in three areas of England. To be eligible, participants needed to have access to the internet and
Centre for Academic Primary
Care, School of Social and
Community Medicine,
University of Bristol, Bristol,
UK (Prof C Salisbury MD,
C Thomas PhD, L Edwards PhD,
S Hollinghurst PhD,
K Garner PhD, M-S Man PhD,
P Dixon PhD); School of Health
and Related Research
email, a Patient Health Questionnaire 9 (PHQ-9) score of at least 10, and a confirmed diagnosis of depression.
Participants were individually assigned (1:1) to either the Healthlines Depression Service plus usual care or usual care
alone. Random assignment was done by use of a web-based automated randomisation system, stratified by site and
minimised by practice and PHQ-9 score. Participants were aware of their allocation, but outcomes were analysed
masked. The Healthlines Service consisted of regular telephone calls from non-clinical, trained health advisers who
followed standardised scripts generated by interactive software. After an initial assessment and goal-setting telephone
call, the advisers called each participant on six occasions over 4 months, and then made up to three more calls at
(ScHARR), University of
intervals of roughly 2 months to provide reinforcement and to detect relapse. Advisers supported participants in the
use of online resources (including computerised cognitive behavioural therapy) and sought to encourage healthier
lifestyles, optimise medication, and improve treatment adherence. The primary outcome was the proportion of
participants responding to the intervention (defined as PHQ-9 <10 and reduction in PHQ-9 of ≥5 points) at 4 months
Sheffield, Sheffield, UK
(Prof A O’Cathain PhD,
Prof J Nicholl DSc, A Foster MPH,
K Horspool MSc); Bristol
Randomised Trials
after randomisation. The primary analysis was based on the intention-to-treat principle without imputation and all
serious adverse events were investigated. This trial is registered with Current Controlled Trials, number
ISRCTN 14172341.
Collaboration, School of Social
and Community Medicine,
University of Bristol, Bristol,
UK (A A Montgomery PhD,
D Gaunt MSc); NHS England
Findings Between July 24, 2012, and July 31, 2013, we recruited 609 participants, randomly assigning 307 to the
South (Wessex), Southampton,
UK (S Large PhD); Department
Healthlines Service plus usual care and 302 to usual care. Primary outcome data were available for 525 (86%)
of Psychology, University of
participants. At 4 months, 68 (27%) of 255 individuals in the intervention group had a treatment response compared
Southampton, Southampton,
with 50 (19%) of 270 individuals in the usual care group (adjusted odds ratio 1·7, 95% CI 1·1–2·5, p=0·019). Compared
UK (Prof L Yardley PhD); Division
with usual care alone, intervention participants reported improvements in anxiety, better access to support and advice,
of Psychiatry, University
College London, London, UK
greater satisfaction with the support they received, and improvements in self-management and health literacy.
(Prof G Lewis PhD); Faculty of
During the trial, 70 adverse events were reported by participants, one of which was related to the intervention
Health Sciences, University of
(increased anxiety from discussing depression) and was not serious.
Southampton, Southampton,
UK (A Rogers PhD, C Pope PhD);
and Nottingham Clinical Trials
Interpretation This telehealth service based on non-clinically trained health advisers supporting patients in use of
Unit, University of
internet resources was both acceptable and effective compared with usual care. Our results provide support for the
Nottingham, Nottingham, UK
development and assessment of similar interventions in other chronic disorders to expand care provision.
(A A Montgomery)
Correspondence to:
Funding National Institute for Health Research (NIHR).
Prof Chris Salisbury, Centre for
Academic Primary Care, School
of Social and Community
Copyright © Salisbury et al. Open Access article distributed under the terms of CC BY.
Medicine, University of Bristol,
Bristol BS8 2PS, UK
Introduction
with health professionals. These technologies, broadly
c.salisbury@bristol.ac.uk
The growing prevalence of chronic disorders is a challenge termed telehealth, include telephone support, messaging,
to the capacity of health-care systems, which are mainly internet-based approaches, and remote monitoring.
based on face-to-face consultations between patients and Some interventions can be automated and supported by
doctors. There is international interest in the potential of non-clinically trained staff, which could expand provision
technology to enable individuals to manage their own and increase access to care. UK health policy envisages
1
health problems, reducing the need to make appointments telehealth becoming mainstream. In the USA, the
www.thelancet.com/psychiatry Vol 3 June 2016
515

Articles
Research in context
Evidence before this study
Some evidence suggested that professional support could
improve the effectiveness of computerised CBT. More recent
reviews reinforce these conclusions. Although there is much
policy enthusiasm for telehealth programmes, few large
pragmatic trials of such programmes have been done with
real-world implementation and a broad range of patients.
There is much scientific literature about specific telehealth
interventions for different chronic disorders. Before developing
the Healthlines intervention, we did a meta-review of existing
systematic reviews. We searched MEDLINE, Embase/AMED,
PsycInfo, Web of Science, Database of Abstracts of Reviews of
Effects (DARE), and The Cochrane Library between Jan 1, 2005,
and March 31, 2010, for relevant systematic reviews. We used
combinations of search terms for systematic reviews, telehealth,
or telemedicine and a list of important chronic disorders
(appendix p 13). We concluded that, although many studies had
been done, many of these were of poor quality, with small
sample sizes, methodological limitations, short-term follow-up,
and insufficient theoretical basis for interventions. This
conclusion was supported by a more recent evidence synthesis
that showed evidence of publication bias and concluded that the
evidence base for telehealth for chronic disorders is “on the
whole weak and contradictory”. We supplemented these reviews
with a meta-review of existing systematic reviews of telehealth
for depression, focusing on which technologies seemed to work
best for which groups of patients. These reviews provided
support for telephone-based and computerised-interventions
for depression, especially internet interventions based on
cognitive behavioural therapy (CBT), although limitations and
inconsistencies in the evidence base were again notable.
Added value of this study
Our trial builds on previous evidence by testing a new model of
care that combines the use of various telehealth tools supported
by non-clinically trained staff. Our study provides robust support
for the potential role of an integrated telehealth approach to
support patients with depression in primary care. Our results
showed that the Healthlines Depression Service increased the
proportion of patients achieving a treatment response and also
led to improvements in anxiety, self-management, patient
satisfaction, and access to health care.
Implications of all the available evidence
Telehealth interventions for chronic disease can lead to small
but meaningful improvements for patients. Basing the
intervention design on a coherent conceptual model and
including multiple evidence-based approaches in combination
is likely to enhance effectiveness. The promising findings from
this study justify developing similar interventions for other
chronic disorders.
Veterans Health Administration has enrolled more than implement those strategies in ways that reduce the
50000 patients in a home telehealth programme, and need for face-to-face contact with clinicians, use of
the Renewing Health Consortium is testing telehealth specific telehealth tools which have the best evidence
programmes in nine European countries.
of effectiveness, and use of technology to ensure
Despite this enthusiasm for telehealth, evidence of integration with primary care. The aim is to improve
benefit is ambiguous and inconsistent, in part due access to care, expand care provision without relying on
to insufficient theoretical underpinning for many clinical staff, deliver more effective treatment, and
2,3
interventions.
Systematic reviews in different improve self-management, leading to improved patient
disorders have shown benefits from specific tech- outcomes.
nologies in selected patients, but few studies have
This paper describes a trial of this approach in the
investigated their effectiveness in large-scale, real-world example of depression. Depression affects 3% of adults
2
implementation. Furthermore, the focus on specific in the UK, is a chronic or relapsing disorder in about half
applications is of limited value because their use cannot of those affected, accounts for about a quarter of primary
be considered in isolation from the health-care system. care consultations, and has substantial adverse effects on
5
Technology-based solutions have frequently been quality of life. There is evidence from systematic reviews
introduced without incorporation into existing models in favour of telehealth interventions for depression,
6
of service provision. The NHS Five Year Forward View including those delivered by telephone or the
7–10
proposes that what is needed is so-called combinatorial internet. However, more than half of patients drop out
9
innovation—an approach that combines different of online therapy, and therapist support alongside
technologies and changed ways of working to transform internet-based interventions increases effectiveness and
4
7,9,10
care delivery.
We developed a new model of care to improve
improves retention of participants.
Through a programme of research, including literature
11
12
management of people with chronic disease based on reviews, qualitative research, and a survey of patient
making the best use of various technologies in views, we developed an evidence-based conceptual
13
combination, supported by non-clinically trained staff model for effective use of telehealth to improve
and integrated with the health-care system. This management of chronic disorders—the TElehealth in
14
See Online for appendix approach incorporates recognised strategies to improve CHronic disease (TECH) model (appendix
p 1).
15
management for chronic disease, use of technology to Building on the Chronic Care Model, the TECH model
516
www.thelancet.com/psychiatry Vol 3 June 2016

Articles
proposes that telehealth programmes need to address either online or by post. If respondents indicated
engagement of patients and health providers, promotion suicidal ideas during assessment of eligibility we asked
of self-management behaviours, optimisation of treat- their GP to assess whether they should be excluded
ment, and care coordination, and they need to be from the trial.
provided in partnership with usual primary care
providers. We used the TECH model to design and Randomisation and masking
assess telehealth interventions (The Healthlines Service) Participants who consented to participate were
for two exemplar common chronic disorders: depression individually randomly allocated (1:1) to either the
and raised cardiovascular risk. In this trial, we aimed to intervention plus usual care group or the usual care
assess the effectiveness of the Healthlines Service plus alone group. Allocation was done with an automated web
usual care compared with usual care alone for patients randomisation system developed independently by the
with depression in a large-scale, real-world application.
Bristol Randomised Controlled Trials Collaboration
(Bristol, UK). Randomisation was stratified by location
(Bristol, Sheffield, or Southampton) and minimised by
general practice and baseline PHQ-9 depression score
Methods
Study design and participants
This study was a pragmatic, multicentre, randomised (10–14, 15–19, ≥20). Participants were notified of their
controlled trial. Participants were recruited in three areas allocation by email. Participants were aware of their
of England (around Bristol, Sheffield, or Southampton). assignment to the intervention or control groups.
The trial was approved by the National Research Ethics Outcome data were collected by participant self-report or
Service Committee South West–Frenchay (Reference electronic download from general practice computer
12/SW/0009). The study protocol has been reported records, and the trial statisticians analysed the outcomes
16
previously. Readers interested in obtaining further masked to allocation.
information about the intervention materials should
refer to the study website.
Procedures
For the study website see
http://www.bristol.ac.uk/
healthlines/
Participants were recruited from 43 general practices, This was a complex intervention incorporating use of
which covered populations with various sociodemo- technologies supported by non-clinically trained staff to
graphic characteristics. Inclusion criteria were a score address each of the components of the TECH model.
17
on the Patient Health Questionnaire 9 (PHQ-9) of at To apply the model to a specific clinical disorder, we
least 10 points and a confirmed diagnosis of depression sought to identify telehealth tools that had some evidence
on the Clinical Interview Schedule-Revised (CIS-R) of effectiveness in that disorder and that could be used to
18
scale, being aged 18 years or older, and having access implement the strategies within the TECH model.
to a telephone, the internet, and email. Participants Therefore the specific telehealth tools varied for the
were excluded if they were currently receiving therapy different exemplar disorders, but were delivered by the
or case management from a mental health worker; had Healthlines Service within a common intervention
given birth in the previous 12 months; had a history of framework (shown for depression in the appendix [p 2]).
major bipolar disorder, psychotic illness, dementia,
The intervention was originally provided by staff from
severe learning disability, or substance dependency; NHS Direct, which, until March 2014, provided various
were receiving palliative care; had a significant suicide telehealth services through a national network of call
risk; or were unable to communicate verbally in centres and a website. After NHS Direct closed in March,
English. Potentially eligible participants were also 2014, the Healthlines Service was transferred to Solent
excluded if their general practitioner (GP) deemed that NHS Trust. The Healthlines Service was based on regular
participation would cause distress (eg, because of a telephone calls from a health adviser who was supported
recent bereavement).
by interactive software. The advisers were not clinically
Potentially eligible participants were identified via qualified but had experience working for NHS Direct
searches of computerised general practice records for and had a further 3 weeks of training in health
patients who had consulted a doctor for depression or coaching, motivational interviewing, antidepressant drug
low mood, or who had been prescribed antidepressants treatment, use of a computerised cognitive behavioural
within the previous 2 months. We asked GPs to screen therapy (CBT) programme, and the Healthlines
the list to remove patients with known exclusion criteria. telephone support software. With the help of scripts
A random sample of potentially eligible patients in each generated by the interactive software, the advisers
practice were sent postal information about the study supported participants in addressing their own health
and a PHQ-9 questionnaire. Patients who expressed an goals and directed them to relevant online resources,
interest in the study and had a PHQ-9 score of at least including reliable health information, interactive pro-
10 were screened by researchers over the telephone to grammes, and relevant apps and widgets (eg, to help
confirm eligibility, including use of the CIS-R to confirm with increasing exercise). The advisers emailed links to
a diagnosis of depression. Eligible patients were asked appropriate websites to participants or sent them the
to complete a baseline questionnaire and consent form, information by post.
www.thelancet.com/psychiatry Vol 3 June 2016
517

Articles
The web portal also provided information about
depression, links to relevant patient-led organisations
and a link to Big White Wall, which is a digital mental
health network that includes a clinically moderated
online forum.
16570 patients invited
14195 excluded
9555 no response
4400 declined to participate
240 other
After an initial assessment and goal-setting telephone
call, the advisers called each participant on six occasions
roughly equally spaced over 4 months, and then made
up to three more calls at roughly two month intervals to
provide reinforcement and to detect relapse. As well as
providing support in use of the CBT programme (online
or in book form), the telephone scripts included modules
covering the monitoring of depression symptoms, drug
treatment, medication adherence, exercise, and alcohol
use. To ensure coordination with primary care, the
advisers sent regular progress reports to participants’
GPs by email and copied them to participants via the
Healthlines portal. In cases of inadequate treatment
response, the advisers contacted participants’ GPs to
2375 assessed for eligibility
1766 excluded
1649 did not meet inclusion criteria
967 PHQ-9 score <10
438 no CIS-R depression diagnosis
67 no internet or email
177 any other reason
25 declined to participate
92 other
609 enrolled
recommend escalation of medication, enclosing
a
5
summary of current treatment guidelines. This
approach was based on our aim to support, rather than
to undermine, the work of the participants’ main
primary care providers. Each participant was telephoned
by the same adviser on each occasion when possible,
since we had identified that it was important to avoid an
anonymous call-centre approach to promote participant
609 randomly assigned
307 assigned to intervention plus usual care
307 received intervention
302 assigned to usual care alone
301 received usual care
86 received in full (9 or 10 calls)
131 received in part (3 to 8 calls)
78 received little or none (2 calls or fewer)
12 missing data
1 did not receive allocated usual care
1 received part of intervention due to
administrative error
11
engagement. Photographs of the advisers were provided
on the Healthlines web portal to enhance the sense of
personal care. To facilitate access to care, the Healthlines
Service was available from 1000 h to 2000 h on weekdays
and from 1000 h to 1400 h on Saturdays.
NHS Direct was closed down towards the end of the
trial, therefore delivery of the intervention was paused
for 2 months while it was transferred to Solent NHS
Trust. Two-fifths of participants had a gap in service
provision during this transfer and some did not receive
the full number of intended phone calls before the end of
their 12 month follow-up period.
255 assessed at 4 months
52 excluded from analysis
1 primary outcome incomplete
27 withdrawn from study
24 lost to follow-up
270 assessed at 4 months
32 excluded from analysis
2 primary outcome incomplete
9 withdrawn from study
21 lost to follow-up
252 assessed at 8 months
61 excluded from analysis
1 primary outcome incomplete
33 withdrawn from study
27 lost to follow-up
263 assessed at 8 months
39 excluded from analysis
17 withdrawn from study
22 lost to follow-up
Outcomes
255 assessed at 12 months
52 excluded from analysis
36 withdrawn from study
16 lost to follow-up
261 assessed at 12 months
41 excluded from analysis
19 withdrawn from study
20 lost to follow-up
2 died
The primary outcome was the proportion of participants
17
responding to the intervention, defined as a PHQ-9
score of less than 10 and a reduction in PHQ-9 of at least
5 points, 4 months after randomisation. Secondary
outcomes were PHQ-9 as a binary outcome at 8 and
12 months and as a continuous outcome at 4, 8, and
Figure 1: Trial profile
Primary outcome incomplete indicates patients who did not answer enough questions in their Patient Health
Questionnaire-9 to allow calculation of a valid score.
21
22
12 months; anxiety (GAD-7); quality of life (EQ-5D-5L);
satisfaction with treatment received and with amount of
For the Big White Wall see
http://www.bigwhitewall.com
Participants were given access to a Healthlines web support received (patient satisfaction); perceived access to
portal, which linked to several resources, including the care; self-management skills and self-efficacy (HeiQ);
23
Living Life to the Full Interactive (LLTTFi) programme. use of telehealth interventions; medication adherence
24
25
This interactive multimedia programme delivers (Morisky); health literacy (eHEALs); and perceptions of
26
computerised CBT-based treatment for depression care coordination (Haggerty). We pre-specified questions
through six self-directed sessions, to be completed about internet use and experience but did not analyse
19
roughly every two weeks. Alternatively, participants these questions in order to reduce the number of
20
could choose to follow similar material in book form.
secondary outcomes. These secondary outcomes were
518
www.thelancet.com/psychiatry Vol 3 June 2016

Articles
chosen to assess the effect of the intervention on the
various components of the underlying TECH model.
At baseline, we also obtained details of socio-demographic
characteristics and current treatment for depression. All
measures were collected through patient questionnaires,
completed online or by post, at baseline and at 4, 8, and
12 months after randomisation. We attempted to follow up
non-respondents by email, telephone, and post.
Participants were offered the option of completing just
the primary outcome during their second reminder,
which could have been by post, email or phone. After
12 months, we collected data from computerised general
practice records about consultations and prescriptions for
antidepressants. We obtained data about use of the
intervention from Healthlines Service records. All
potential serious adverse events were investigated.
Adverse events were mainly detected as part of the
collection of data with respect to resource use during the
12 month follow-up period, either collected within
the patient questionnaires or during the review of
participants’ medical records in primary care at the end of
the trial. We collected data about any hospital attendance
or new or recurrent serious medical events recorded in
primary care records during the 12 month period.
Usual care alone
(n=302)*
Intervention plus usual
care (n=307)*
Demographic
Age (years)
50·0 (12·8)†
204/302 (68%)
292/301 (97%)
49·1 (12·9)‡
213/307 (69%)
300/306 (98%)
Female
White
Employment situation
Full-time employment
92/299 (31%)
39/299 (13%)
2/299 (1%)
88/303 (29%)
56/303 (18%)
5/303 (2%)
Part-time employment
Full-time education
Unemployed
13/299 (4%)
78/299 (26%)
2/299 (1%)
14/303 (5%)
73/303 (24%)
4/303 (1%)
Unable to work due to long-term illness/disability
Unable to work due to carer responsibilities
Fully retired from work
Looking after the home
Doing something else
44/299 (15%)
10/299 (3%)
19/299 (6%)
40/303 (13%)
13/303 (4%)
10/303 (3%)
Occupation (most recent or current)
Administrative or secretarial occupations
Associate professional or technical occupations
Elementary occupations
Managers or senior officials
Personal services
51/262 (19%)
37/262 (14%)
21/262 (08%)
32/262 (12%)
28/262 (11%)
11/262 (4%)
35/262 (13%)
35/262 (13%)
12/262 (5%)
49/270 (18%)
35/270 (13%)
19/270 (7%)
41/270 (15%)
27/270 (1%)
15/270 (6%)
42/270 (16%)
29/270 (11%)
13/270 (5%)
Process, plant, and machine operatives
Professionals
Statistical analysis
Sales and customer services
Skilled trades
We chose the sample size pragmatically, taking into
account the size of effect that would be likely to affect
practice and that might be feasible. Assuming a 30%
response to treatment in the control group and 20%
attrition, 300 participants in each group would provide
80% power (α of 5%) to detect a difference of 13% and
90% power (α of 1% ) to detect a difference of 18%.
We used descriptive statistics to compare baseline
characteristics of trial participants by allocated group.
If we detected important imbalances between the
intervention and control groups, these variables were
included in regression models in sensitivity analyses of
the primary outcome. We fitted all regression models
with maximum likelihood estimation, apart from the
regression models that used multiply imputed data, for
which we used QR decomposition.
Highest education qualification achieved
Degree or higher degree
A levels or equivalent
84/298 (28%)
54/298 (18%)
119/298 (40%)
41/298 (14%)
68/303 (22%)
63/303 (21%)
130/303 (43%)
42/303 (14%)
GCSEs/O levels or equivalent
No qualifications
Accommodation
Own accommodation or buying with mortgage
Part-rent or rent accommodation
Live rent free
162/300 (54%)
124/300 (41%)
14/300 (5%)
18·0 (13·0)§
179/307 (58%)
118/307 (38%)
10/307 (3%)
Index of multiple deprivation
Clinical data
18·3 (12·8)‡
Previously treated for depression
PHQ-9 score
258/276 (93%)
16·7 (4·7)†
269/295 (91%)
17·1 (4·5)‡
GAD-7 score
12·4 (5·0)¶
13·5 (4·6)||
Categorised with mild depression from CIS-R
52/302 (17%)
39/307 (13%)
165/307 (54%)
103/307 (34%)
251/289 (87%)
We did the primary analysis with a mixed-effects logistic
regression model adjusted by the stratification and
minimisation variables of site (categorical), baseline value
of the outcome (continuous), and general practice
(categorical, included as a random effect). In the primary
analysis, we included all participants with complete data in
accordance with their allocated treatment, based on the
principle of intention to treat without imputation. We
investigated the robustness of the primary analysis to
different methods of imputing missing outcome data:
Categorised with moderate depression from CIS-R 148/302 (49%)
Categorised with severe depression from CIS-R
Taking antidepressants
102/302 (34%)
258/288 (90%)
Data are n/N (%) or mean (SD). GCSE=General Certificate of Secondary Education. PHQ-9=Patient Health
Questionnaire 9. GAD-7=Generalised Anxiety Disorder scale. CIS-R=Clinical Interview Schedule-Revised scale.
*Denominators vary because of missing data. †n=302. ‡n=307. §n=301. ¶n=298. ||n=304.
Table 1: Baseline characteristics
simple imputation that assumed no treatment response or
By fitting interaction terms between trial groups and
27
multiple imputation (appendix p 5). We also investigated subgroup variables, we investigated whether any effect of
the effect of omitting GP practice as a random effect, and the Healthlines intervention differed by subgroups
adjusting for time between randomisation and follow-up defined at baseline by sex, age, PHQ-9 score, severity of
and variables imbalanced between the groups at baseline.
depression, or use of antidepressant medication.
www.thelancet.com/psychiatry Vol 3 June 2016
519

Articles
To estimate the effect of the Healthlines intervention at full access to all the data in the study and had
12 months when received as intended, we described final responsibility for the decision to submit for
compliance as little or none (two telephone calls or fewer), publication.
partial (three to eight calls), or full (nine or ten calls). We
estimated the complier-average causal effect (CACE) with Results
principal stratification in two ways, classifying partial Participants were recruited and randomly assigned to
compliers as either non-compliers or full compliers.
study groups between July 24, 2012, and July 31, 2013.
We did repeated measures analyses of the primary Of 2375 participants assessed for eligibility, 1649 (69%) did
outcome, as both a binary and continuous outcome, not meet eligibility criteria, mainly because they did not
using appropriate mixed-effect regression models, meet the diagnostic criteria for depression (1405 [85%] of
including general practice as
a random effect, 1649 patients not meeting the eligibility criteria). Of the
participants as random intercepts, and assuming an 726 eligible participants, 609 participants were randomly
independent covariance structure. We investigated assigned, with 307 assigned to intervention plus usual care
whether the between-group difference changed over and 302 assigned to usual care alone (figure 1). Primary
time by fitting an interaction between trial group and outcome data were obtained for 255 (83%) participants in
follow-up timepoint. In the absence of any such the intervention group and 270 (89%) in the usual care
interactions, we report the average between-group effect group. We retained 516 (85%) participants until the final
across 12 months.
12 month follow-up assessment.
Secondary outcomes were analysed in a similar manner
Table 1 shows baseline characteristics of participants
to the primary analysis. We estimated between-group who entered randomisation. Most participants had
effects with linear, logistic, or binomial mixed-effect severe and enduring depression. Of patients with data
regression models, adjusted for stratification and available, 205 (34%) were classified as severely depressed
minimisation variables and continuous value of the at baseline, 527 (92%) had previously been treated for
outcome at baseline (if appropriate). Participants were depression, and 509 (88%) were currently taking
analysed as randomised, without imputation of missing antidepressant medication. The intervention and
data. We did no sensitivity analyses for secondary control groups were mostly well balanced, although
outcomes. To reduce the number of statistical comparisons, there were small differences in work status, education,
we estimated between-group differences for secondary accommodation, depression severity, and use of
outcomes only at the 12 month follow-up timepoint. We antidepressants.
described serious adverse events by study group.
Analysis of the primary outcome showed that 68 (27%)
We did analyses using Stata version 13.1. The trial of 255 participants in the intervention group responded
was overseen by an Independent Trial Steering to treatment at 4 months, compared with 50 (19%) of
Committee and an Independent Data Monitoring 270 participants in the control group (adjusted odds ratio
Committee, who agreed a statistical analysis plan [OR] 1·7, 95% CI 1·1–2·5, p=0·019; table 2). This 8%
before analysis. Two patient and public representatives difference between groups in the proportion of
actively contributed as members of the Trial Steering participants responding to treatment is equivalent to a
Committee. The trial is registered with Current number needed to treat of 12. This result was robust to
Controlled Trials, number ISRCTN 14172341.
sensitivity analyses (appendix p 6). No evidence suggested
that the intervention was differentially effective for any
subgroups defined by baseline characteristics.
Role of the funding source
The funder of the study had no role in study design,
The proportion of participants who responded to
data collection, data analysis, data interpretation, or treatment increased in both arms at 8 and 12 months
writing of the report. The corresponding author had (table 2). Although the between-group difference seemed
to diminish with increasing duration of follow-up, there
was no evidence of an interaction between treatment
Usual care
Intervention plus Adjusted odds ratio
p value
group and timepoint (pinteraction=0·402). Therefore, we
estimated an overall treatment effect from all follow-up
data at 4, 8, and 12 months in two repeated measures
analyses, binary and continuous. This suggested a positive
average effect of the intervention over the 12 month
follow-up period (binary response to treatment p=0·035).
Mean PHQ-9 scores also decreased over time for both
groups (table 3), with no evidence of an interaction
between treatment group and timepoint (pinteraction=0·345).
A repeated measures analysis of continuous PHQ-9
scores provided evidence of a small overall treatment
effect (p=0·045).
usual care
(95% CI)
4 months
50/270 (19%)
61/263 (23%)
86/261 (33%)
··
68/255 (27%)
75/252 (30%)
95/255 (37%)
··
1·7 (1·1–2·5)
1·4 (1·0–2·2)
1·2 (0·9–1·8)
1·6 (1·0–2·6)
0·019
NC
8 months
12 months
NC
Average of 4, 8, and
12 month effects*
0·035
Data are n/N (%) unless otherwise stated. All analyses are adjusted by site (Bristol, Sheffield, or Southampton) and
baseline PHQ-9 score. GP practice is included as a random effect. NC=not calculated at individual timepoints; analysis
plan specified repeated measures analysis. *Based on a repeated measures analysis that was additionally adjusted by
follow-up timepoint as a categorical variable.
Table 2: Repeated measures analysis of treatment response as a binary outcome
520
www.thelancet.com/psychiatry Vol 3 June 2016

Articles
Usual care
Intervention plus usual care
Adjusted difference in
means (95% CI)
p value
Patients with
data available
Unadjusted
mean (SD)
Patients with
data available
Unadjusted mean
(SD)
PHQ-9 score (continuous outcome)
Baseline
302
270
263
261
··
16·7 (4·7)
13·8 (6·2)
13·4 (6·2)
11·9 (6·4)
··
307
255
252
255
··
17·1 (4·5)
13·3 (6·1)
12·4 (6·2)
11·6 (6·2)
··
··
NA
4 months
–0·8 (–1·8 to 0·1)
–1·2 (–2·2 to –0·3)
–0·5 (–1·5 to 0·5)
–0·8 (–1·6 to 0·0)
–1·1 (–2·0 to –0·2)
–0·003 (–0·04 to 0·04)
0·5 (0·3 to 0·6)
0·3 (0·0 to 0·6)
0·6 (0·4 to 0·7)
··
NC
8 months
NC
12 months
NC
Average of 4, 8, and 12 month effects*
0·045
0·019
0·896
<0·001
0·038
<0·001
21
Generalised anxiety (GAD-7)
237
227
184
232
177
235
237
238
239
238
179
235
··
9·2 (5·8)
0·564 (0·30)
3·3 (0·9)
4·2 (1·9)
2·1 (0·9)
2·4 (0·9)
2·4 (0·9)
2·6 (0·6)
2·6 (0·5)
2·8 (0·6)
3·4 (0·9)
3·7 (0·8)
··
223
219
193
216
185
221
221
221
221
220
173
220
174
179
219
210
··
8·7 (5·5)
0·569 (0·30)
3·7 (0·9)
4·5 (1·9)
2·6 (0·8)
2·5 (0·9)
3·0 (0·5)
2·7 (0·6)
2·8 (0·5)
2·9 (0·6)
3·2 (1·1)
3·9 (0·8)
··
22
Quality of life (EQ-5D-5L)
Satisfaction with treatment†‡
Difficulties with obtaining access to care†‡
Satisfaction with amount of support received†‡
23
Self-management skills and self-efficacy (heiQ)
Physical activity† (heiQ health-directed behaviour scale)
Self-monitoring and insight†
0·1 (<0·0 to 0·2)
0·1 (0·0 to 0·2)
0·0 (–0·1 to 0·1)
0·2 (0·1 to 0·2)
0·2 (0·1 to 0·3)
–0·1 (–0·2 to 0·1)
0·2 (0·1 to 0·4)
··
0·118
0·005
0·480
0·001
<0·001
0·511
Constructive attitudes and approaches†
Skill and technique acquisition†
Health services navigation†
24
Adherence to antidepressant medication (Morisky) †
25
Health literacy (eHEALs) †
<0·001
26
Care coordination (Haggerty)
176
236
230
··
3·1 (2·2)
3·1 (1·0)
3·2 (1·2)
··
3·5 (2·4)
3·2 (1·0)
3·1 (1·2)
··
Role clarity and coordination†
–0·1 (–0·2 to 0·1)
0·3 (–0·1 to 0·8)
0·1 (–0·1 to 0·3)
0·0 (–0·2 to 0·2)
0·361
0·173
0·247
0·841
Evidence of a care plan†
Overall experience of organisation of health care†
Self-organisation of health care†
··
··
··
··
All analyses are adjusted by site (Bristol, Sheffield or Southampton), baseline outcome (if measured) and baseline PHQ-9 score. GP practice is included as a random effect. PHQ=patient health questionnaire.
HeiQ=Health Education Impact Questionnaire. NC=not calculated at individual timepoints; analysis plan specified repeated measures analysis. NA=not applicable. *Based on a repeated measures analysis that was
additionally adjusted by follow-up timepoint as a categorical variable. †Higher score is more positive (fewer access difficulties, greater satisfaction). ‡Based on scales generated before the main trial analysis using
principal components analysis and incorporating questions taken from existing validated questionnaires or constructed for this research.
Table 3: Secondary outcomes and measures used to assess them
Table 3 shows the findings for the secondary outcomes,
except for use of telehealth interventions, which is shown in
65
60
the appendix (p 7). The intervention was associated with
55
50
improvements in anxiety, as assessed with the GAD-7
21
instrument. It was also associated with improvements in
40
35
30
25
20
15
10
several measures of self-management skills and attitudes
25
and improved health literacy. Participants in the
intervention group expressed greater satisfaction with
access to health care, treatment, and amount of support
they received. There was no evidence of improved adherence
with antidepressant medication or improved care
coordination (table 3) or that participants made more use of
5
0
0
1
2
3
4
5
6
7
8
9
10
other health-related technologies (appendix p 7). For ease of
presentation, table 3 only shows data about secondary
outcomes after 12 months’ follow-up. Details of findings
after 4 and 8 months are available in the appendix (p 8).
Use of the intervention varied considerably, with some
Number of encounters
Figure 2: Distribution of Healthlines telephone encounters from 295 patients
Data about use of the intervention were missing for 12 participants who received
the intervention.
participants dropping out after a few calls while others ten encounters. The median number of encounters was
completed the complete course of telephone calls. five (IQR 2–9, 295 participants had available data), mean
Figure 2 shows the number of Healthlines encounters encounter duration was 18·5 min (SD 12·7), and there
with intervention participants out of a maximum of was a total of 1972 encounter calls.
www.thelancet.com/psychiatry Vol 3 June 2016
521

Articles
PHQ-9 response at 12-month follow-up
Partial compliers
classified as
non-compliers
Partial compliers
classified as full
compliers
Usual care
Intervention plus
usual care
No intervention received (0–2 encounters started)
Partial intervention received (3–8 encounters started)
Full intervention received (9–10 encounters started)
86/261 (33%)
16/52 (31%)
42/111 (38%)
36/86 (42%)
··
··
··
··
··
··
1·9 (0·99–3·48)
1·3 (0·87–1·95)
Data are n/N (%) or unadjusted odds ratio (95% CI). We estimated the effect among compliers at 12 months because this was the scheduled duration to receive the entire
intervention. Three participants who never received the Healthlines intervention and one participant who only received an unscheduled non-encounter call are categorised as
receiving none of the intervention. 12 participants had missing encounter data.
Table 4: Complier-average causal effect analysis at 12 months
Only 86 (28%) of 307 of participants in the intervention Discussion
group received at least nine of the potential maximum This report describes a large pragmatic trial of the
of ten telephone encounters. Table 4 shows the crude Healthlines depression service, a novel approach to the
estimated effect at 12 months among compliers, after delivery of health care through a combination of various
classifying partial compliers in the intervention group as technologies to improve treatment and self-management,
either non-compliers or full compliers. The results supported by lay staff, and integrated with the health-care
suggest an increase in effect when compliance is defined system. The intervention was clinically effective, leading
as completion of most or all of the Healthlines encounters. to improved response to treatment at 4 months’ for
Use of different components of the intervention also participants receiving the intervention plus usual care
varied. Generally, sustained usage was low, although compared with those receiving usual care alone, and the
some individuals used the resources extensively. Out of size of the effect is important, with a number needed to
285 participants in the intervention group with valid data, treat of 12. Furthermore, compared with participants
237 chose to use the online computerised CBT programme who received usual care alone, participants who received
(LLTFi) and 53 chose to follow the book (eight individuals the intervention reported reduced anxiety, improved
used both). Of those choosing LLTFi, 204 participants access to health support and advice, greater satisfaction
logged on to the website and completed a median of two with the treatment and the amount of help they
(IQR 0–5) of the seven modules. 66 participants who received, and improvements in self-management
logged on did not complete the first module. attitudes and skills.
98 participants registered with Big White Wall and logged
The effect of the intervention was greatest after
on to the website at least once. The median number of 4 months of follow-up, and the difference between the
logins was two (range 1–104), the median number of intervention and usual care groups narrowed with time
interactions (posts) was zero (range 0–49), and the time as both groups improved. After 12 months of follow-up,
spent on the site was 0·30 h (range 0·00–11·17).
evidence suggested no benefit from the intervention,
In a prespecified exploratory analysis of the process of although an average benefit existed for the whole
care, participants in the intervention group had slightly 12 month follow-up period. Therefore, the main effect of
more primary care consultations during 12 months the intervention in terms of depression was increased
(mean 15·4, SD 11·0) than did those in the control group speed of recovery, which is nevertheless important.
(mean 14·3, SD 12·5; adjusted incidence ratio 1·07, However, only a few participants fulfilled our criteria for
95% CI 0·96–1·19, p=0·22). There were no differences response to treatment, whether they received the
between the intervention and control groups in the intervention or usual care, and the mean PHQ-9 score
proportion of participants who had been prescribed an was more than 10 in both groups at all timepoints,
antidepressant during the trial, the number of changes suggesting substantial residual symptoms. Notably,
in antidepressant type or dose, or the number of many participants had severe and enduring depression
participants who were taking an antidepressant at when they were recruited to the trial, most were taking
follow-up (appendix p 11).
antidepressants, and most remained symptomatic at the
During the course of the trial, 70 adverse events were end of the trial. These results suggest that many
reported by participants (36 in the usual care group and participants had so-called treatment-resistant depression,
34 in the intervention group), of which one was related which is unsurprising, since this pattern of a long history
to the intervention (increased anxiety from talking of depression and poor treatment response is typical
about depression) and was not serious (appendix p 12). of patients managed in primary care. More than
Two patients died (one due to COPD and one due to three-quarters of patients treated with antidepressants in
throat cancer), both in the usual care group.
primary care for at least 6 weeks have clinically significant
522
www.thelancet.com/psychiatry Vol 3 June 2016

Articles
residual symptoms because of complex reasons such as
The most similar previous study, which was done in
insufficient effect of or intolerance to medication, non- the USA, was based on a so-called collaborative care
28
adherence, and undertreatment. This evidence is model. Care managers were provided with a web-based
consistent with the notion that depression is often a clinical decision-support system, but the intervention did
30
chronic disease and shows the challenge of improving not make use of any other telehealth approaches.
care for patients with treatment-resistant depression.
The adjusted odds ratio for treatment response was
Management options for treatment-resistant depression 1·9 (95% CI 1·09–3·45) at 6 months follow-up, similar to
include changes of medication, increased dosage, the estimate of treatment response at 4 months in this
addressing of non-adherence, or addition of psychological UK-based Healthlines trial (1·7, 95% CI 1·1–2·5). Several
therapy. This Healthlines intervention was designed to studies of other collaborative care interventions have
address all of these issues, for example through strategies been done that often involve telephone follow-up from
to promote adherence, offering computerised CBT, and case managers, but have not included the integrated use
sending guidance to GPs about changes to medication. of internet or other telehealth resources, which is a
Our data suggest that the proposed mechanisms of action defining characteristic of the Healthlines intervention.
31
were only partly achieved, with no evidence of improved A Cochrane review estimated that the risk ratio for
medication adherence or that GPs were more active in response from collaborative care for depression was
changing medication. Many participants only used some 1·32 (95% CI 1·22–1·43) in the short term (0–6 months)
components of the intervention to a small extent, and only and 1·31 (95% CI 1·17–1·48) in the medium term
a few received the full course of telephone consultations (6–12 months). The findings from our trial are similar to
that formed the core of the intervention. However, some these results, but slightly more positive (the observed
participants engaged with the intervention extensively, adjusted odds ratio at 4 months follow-up equates to a
and our analysis suggested that adherence to the risk ratio of 1·5, 95% CI 1·1–1·9).
intervention was associated with enhanced effectiveness.
This trial has several strengths, including a large
Therefore, it is likely that the main mechanisms of sample size, inclusion of participants with a confirmed
effect were the support and motivation received from diagnosis of depression, the high proportion of
the Healthlines advisers to improve self-management, participants retained, and recruitment from multiple
along with the addition of computerised CBT to sites. The study is highly pragmatic, testing a new
existing drug treatment. To improve effectiveness of approach to care based on telehealth, which could be
the intervention, exploration of ways to improve quickly and easily rolled out nationally. The intervention
communication with primary care will be necessary to was carefully developed with an evidence-based
improve medication optimisation. Such exploration conceptual model that helps us to explore causal
might be easier outside of a research context once mechanisms and provides a framework for development
the intervention is well established and has been and evaluation. Alongside the trial, we did parallel
normalised within the health-care system. To improve process and economic assessments that will be reported
patient engagement, prescreening might be beneficial separately.
to target the intervention at those who would be most
The study also has several limitations. First, only a small
proportion of those patients sent information about the
interested in using this approach.
6–10
Findings from systematic reviews have suggested trial expressed interest in participating. Targeting of the
positive effects on patient outcomes from various initial mailing to the population of patients meeting our
specific telehealth interventions for depression, severity criteria for depression was not possible. Therefore,
although some investigators have pointed out we cannot establish how many of the individuals who did
methodological limitations in the evidence base, not respond to the invitation would not have been eligible
including the abundance of small feasibility studies, because they did not have depression or would not meet
assessment of a small range of outcomes, and only severity criteria and how many were not interested in the
8
short-term follow-up. Another systematic review
intervention or did not wish to participate in this research
described the evidence for internet-supported treatment study. Absence of interest or ability to use the intervention
for depression as promising but inconclusive. Several is not necessarily a threat to external validity, because we
trials of computerised CBT for depression have had recognise that telehealth interventions are not necessarily
positive results, but these have had limitations and a suitable or of interest to everyone. However, if those
high quality trial of two computerised CBT packages responding to the invitation to participate in the research
with regular telephone support showed no evidence of are unrepresentative, the findings might not be
benefit from either package compared with usual GP generalisable. If the intervention was implemented widely,
29
care. Very few pragmatic trials have been done to study it is possible that people with different characteristics or
system-level telehealth interventions for depression that severity of depression would take it up, but whether the
integrate various technologies and new ways of working benefit would be greater or less than that we observed is
implemented by non-clinically trained staff. Our study impossible to know. This problem of low recruitment is
helps to fill this evidence gap.
common to other large trials of depression in primary
www.thelancet.com/psychiatry Vol 3 June 2016
523

Articles
care, which have recruited patients by use of similar Direct. AAM supervised the statistical analysis. All authors contributed
28,32
to decision making throughout data collection, analysis, interpretation,
and reporting and approved the final version of this manuscript.
methods.
In this trial, once eligibility had been
confirmed, the proportion of eligible individuals who
agreed to participate was high (84%).
Another limitation is the high proportion of participants
who received less than a full course of encounters with
the health adviser. This problem is also common in other
Declaration of interests
CS, GL, AO’C, and JN act as members of various boards for the
National Institute for Health Research (NIHR) but were not on the
board that commissioned this project. The other authors declare no
competing interests.
9
telehealth interventions for depression. In this trial, the
Acknowledgments
This report summarises independent research funded by the NIHR under
its Programme Grant for Applied Research (Grant Reference Number
RP-PG-0108-10011). The views and opinions expressed in this report are
those of the authors and do not necessarily reflect those of the NIHR, the
NHS or the Department of Health. We thank all the patients, health care
professionals, Health Information Advisers, and other NHS Direct staff
who contributed time and effort to make this trial possible. We are grateful
to Dr Sarah Williams at Solent NHS Trust who took over responsibility for
hosting the intervention when NHS Direct closed and Steve Bellerby, who
managed implementation of the Healthlines software at both Trusts. We
thank the members of the Trial Steering and Data Monitoring Committees,
particularly Michelle McPhail and Anne Jacob who were Patient and Public
representatives. We thank the Big White Wall and Living Life to the Full
Interactive for use of these programmes, as well as providing anonymised
data on participant usage. We acknowledge Tom Fahey and
Simon Brownsell who were applicants on the research programme grant
but not directly involved in this trial. We thank Chris Williams for advising
about intervention development and contributing to the training of the
advisers. We are grateful to the Primary Care Research Network (now,
NIHR Clinical Research Network) for assisting us with general practice
recruitment. We wish to acknowledge the work of Frederika Collihole,
Richard Campbell, Ben Davies, Lorna Duncan, Diane Beck, and
Janet Cooke who supported participant recruitment, data entry and trial
administration. This study was designed and delivered in collaboration
with Bristol Randomised Trials Collaboration, a UKCRC Registered Clinical
Trials Unit in receipt of NIHR Clinical Trials Unit support funding.
Copyright of the Healthlines telephone algorithms is held by Solent NHS
Trust. The Living Life to the Full Interactive programme (copyright Media
Innovations Limited) and the Big White Wall (copyright BigWhiteWall Ltd)
missed encounters might have occurred for various
reasons; some individuals chose not to continue with the
telephone calls (possibly because they got better), some
became uncontactable, and others missed encounters
because of the closure of NHS Direct. These findings
might help to explain the positive benefit seen when
response to treatment (based on the PHQ-9 score) was
treated as a binary outcome, yet minimal change seen
when the PHQ-9 score was treated as continuous. This
result suggests that a proportion of individuals benefited
from the Healthlines service, but many people did not,
which would be consistent with the finding that one
group of individuals dropped out of the service after just
two or three encounters and another group completed
the course and with the finding that those who completed
the course gained most benefit. A further limitation
was that, although the overall retention was high, the
retention rate differed slightly between the intervention
and usual care arms. However, multiple imputation of
missing primary outcome data had no effect on the
findings, so we think that it is unlikely that differential
attrition could explain our results.
Our findings show that it is feasible and effective to
provide a scalable intervention for depression delivered by were used with permission.
non-clinically trained advisers working with computerised
algorithms and encouraging people to make use of the
wide range of help available from the internet. Such an
intervention makes it possible to substantially expand
provision of care without being limited by the availability
of clinically trained staff, which would help to meet the
pressing need to expand services for common mental
health problems. Although the absolute benefit from this
intervention in terms of clinical outcomes is small, the
number of patients with the potential to benefit is very
large and the results are sufficiently promising to justify
further development of the intervention to improve its
acceptability and effectiveness. Further research should
explore the benefits of interventions based on a similar
model for other chronic disorders.
References
1
NHS England. Technology Enabled Care Services (TECS). 2015.
http://www.england.nhs.uk/ourwork/qual-clin-lead/tecs/ (accessed
April 14, 2015.
2
3
Wootton R. Twenty years of telemedicine in chronic disease
management—an evidence synthesis. J Telemed Telecare 2012;
18: 211–20.
Flodgren G, Rachas A, Farmer Andrew J, Inzitari M, Shepperd S.
Interactive telemedicine: effects on professional practice and health
care outcomes. Cochrane Database Syst Rev, 2015.
http://onlinelibrary.wiley.com/doi/10.1002/14651858.CD002098.
pub2/abstract (accessed Dec 30, 2015).
4
Five Year Forward View 2014. https://www.england.nhs.uk/
wp-content/uploads/2014/10/5yfv-web.pdf (accessed Dec 30, 2015).
5
National Institute for Health and Care Excellence. Depression in
adults: the treatment and management of depression in adults.
CG902009. http://www.nice.org.uk/CG90 (accessed Oct 8, 2015).
6
7
Mohr DC, Vella L, Hart S, Heckman T, Simon G. The effect of
telephone-administered psychotherapy on symptoms of depression
and attrition: a meta-analysis. Clin Psychol (New York) 2008;
15: 243–53.
Andersson G, Cuijpers P. Internet-based and other computerized
psychological treatments for adult depression: a meta-analysis.
Cogn Behav Ther 2009; 38: 196–205.
Contributors
CS, AO’C, SH, JN, SL, LY, GL, AR, CP, and AAM developed the protocol
for the study, obtained funding, provided methodological advice and
supervised the conduct of the trial. CS led protocol development and the
funding application, acted as chief investigator with overall responsibility
for the conduct of the trial, and led the drafting of the Article.
AO’C supervised the conduct of the trial in Sheffield. CT, M-SM, and LE
acted as trial managers, coordinating the conduct of the trial across the
centres. LE, AF, KG, and KH recruited patients and did the follow-up,
data collection, and data entry. DG developed the statistical analysis plan
and did the statistical analysis. PD contributed to the statistical analysis.
SL coordinated development and delivery of the intervention with NHS
8
9
García-Lizana F, Muñoz-Mayorga I. Telemedicine for depression:
a systematic review. Perspect Psychiatr Care 2010; 46: 119–26.
Richards D, Richardson T. Computer-based psychological
treatments for depression: a systematic review and meta-analysis.
Clin Psychol Rev 2012; 32: 329–42.
10 Saddichha S, Al-Desouki M, Lamia A, Linden IA, Krausz M.
Online interventions for depression and anxiety—a systematic
review. Health Psychol Behav Med 2014; 2: 841–81.
524
www.thelancet.com/psychiatry Vol 3 June 2016

Articles
11 Vassilev I, Rowsell A, Pope C, et al. Assessing the implementability
of telehealth interventions for self-management support:
a realist review. Implement Sci 2015; 10: 59.
12 Segar J, Rogers A, Salisbury C, Thomas C. Roles and identities in
transition: boundaries of work and inter-professional relationships
at the interface between telehealth and primary care.
23 Osborne RH, Elsworth GR, Whitfield K. The Health Education
Impact Questionnaire (heiQ): an outcomes and evaluation measure
for patient education and self-management interventions for people
with chronic conditions. Patient Educ Couns 2007; 66: 192–201.
24 Morisky DE, Green LW, Levine DM. Concurrent and predictive
validity of a self-reported measure of medication adherence.
Med Care 1986; 24: 67–74.
Health Soc Care Community 2013; 21: 606–13.
13 Edwards L, Thomas C, Gregory A, et al. Are people with chronic
diseases interested in using telehealth? A cross-sectional postal
survey. J Med Internet Res 2014; 16: e123.
14 Salisbury C, Thomas C, O’Cathain A, et al. TElehealth in CHronic
disease: mixed-methods study to develop the TECH conceptual
model for intervention design and evaluation. BMJ Open 2015;
5: e006448.
25 Norman DC, Skinner AH. eHEALS: The eHealth Literacy
Scale. J Med Internet Res 2006; 8: e27.
26 Haggerty JL, Roberge D, Freeman GK, Beaulieu C, Breton M.
Validation of a generic measure of continuity of care: when patients
encounter several clinicians. Ann Fam Med 2012; 10: 443–51.
27 Altman DG. Missing outcomes in randomized trials: addressing the
dilemma. Open Med 2009; 3: e51–53.
15 Wagner EH, Austin BT, Von Korff M. Improving outcomes in
28 Wiles N, Thomas L, Abel A, et al. Clinical effectiveness and
cost-effectiveness of cognitive behavioural therapy as an adjunct to
pharmacotherapy for treatment-resistant depression in primary
care: the CoBalT randomised controlled trial. Health Technol Assess
2014; 18.
29 Gilbody S, Littlewood E, Hewitt C, et al. Computerised cognitive
behaviour therapy (cCBT) as treatment for depression in primary
care (REEACT trial): large scale pragmatic randomised controlled
trial. BMJ 2015; 351: h5627.
30 Fortney JC, Pyne JM, Edlund MJ, et al. A randomized trial of
telemedicine-based collaborative care for depression.
J Gen Intern Med 2007; 22: 1086–93.
31 Archer J, Bower P, Gilbody S, et al. Collaborative care for depression
and anxiety problems. Cochrane Database Syst Rev 2012;
10: CD006525.
chronic illness. Manag Care Q 1996; 4: 12–25.
16 Thomas CL, Man M-S, O’Cathain A, et al. Effectiveness and
cost-effectiveness of a telehealth intervention to support the
management of long-term conditions: study protocol for two linked
randomized controlled trials. Trials 2014; 15: 36.
17 Kroenke K, Spitzer R, Williams JW. The PHQ-9 validity of a brief
depression severity measure. J Gen Intern Med 2001; 16: 606–13.
18 Lewis G, Pelosi AJ, Araya R, Dunn G. Measuring psychiatric
disorder in the community: a standardized assessment for use by
lay interviewers. Psychol Med 1992; 22: 465–86.
19 Williams C, Wilson P, Morrison J, et al. Guided self-help cognitive
behavioural therapy for depression in primary care: a randomised
controlled trial. PLoS One 2013; 8: e52735.
20 Williams C. Overcoming depression and low mood: a five areas
approach. 3rd edn. London, UK: Hodder Arnold, 2009.
21 Spitzer RL, Kroenke K, Williams JBW, Lowe B. A brief measure for
assessing generalized anxiety disorder: The GAD-7. Arch Intern Med
2006; 166: 1092–97.
32 Richards DA, Hill JJ, Gask L, et al. Clinical effectiveness of
collaborative care for depression in UK primary care (CADET):
cluster randomised controlled trial. BMJ 2013; 347: f4913.
22 EuroQol Group. EQ-5D-5L User Guide. http://www.euroqol.org/
fileadmin/user_upload/Documenten/PDF/Folders_Flyers/EQ-5D-
5L_UserGuide_2015.pdf (accessed March 31, 2016).
www.thelancet.com/psychiatry Vol 3 June 2016
525