a randomised controlled trial (MATISSE). - ResearchGate

DOI: 10.3310/hta16080
Health Technology Assessment 2012; Vol. 16: No. 8
self-reported well-being was moderate/good and participants reported being generally satisfied
with the care that they received.
Comparison of BSL characteristics across the three study arms indicates that the groups were
well balanced.
Flow of participants through the trial
The proportion of participants who dropped out was less than predicted. The CONSORT
(Consolidated Standards of Reporting Trials) diagram (Figure 1) summarises the assessments
completed at each time point: 361 (87%) were followed up at 12 months and 355 (85%) were
followed up at 24 months. Seven participants died during the follow-up period: two were in the
art therapy arm of the trial, three in the activity group arm and two in the standard care arm of
the trial. Four of the seven deaths were from suicide/probable suicide. Three additional serious
adverse events were reported, one a near-fatal episode of deliberate self-harm and two involving
harm to others. None appeared to be related to the interventions being examined in the study.
Of the 62 (15%) participants who were not followed up at 24 months, 18 (7%) formally withdrew
from the study and seven (2%) died. The remainder either could not be traced or did not take
up repeated offers to be assessed (6%). The attrition rate was similar across arms, and there was
no difference in reasons for attrition (death withdrawal, lost to follow-up). Participants who
completed follow-up had BSL characteristics similar to those who did not (Table 3). However, the
rate of attrition was higher in north London than in other study centres (p < 0.001).
Masking of researchers conducting follow-up interviews
Researchers reported 16 occasions when they became aware of a participants’ allocation status.
On nine occasions this was when clinical staff revealed which arm of the trial the participant was
in, and on seven occasions it was when the participants told the researcher to which arm of the
trial they had been allocated. When researchers attempted to guess allocation status following
completion of the final follow-up interview, approximately half of their guesses were correct
(n = 119, 48%).
Uptake of allocated treatments
The number and proportion of those allocated to art therapy and activity groups are presented
in Table 4. The median delay between randomisation and a person attending their first group
was 61 days for art therapy and 61.5 days for an activity group. Those allocated to art therapy
attended between 0 and 51 groups, whereas those allocated to control groups attended between 0
and 45 groups. Eighty-six (61%) of those randomised to art therapy attended at least one session,
compared with 73 (52%) of those randomised to activity groups (p = 0.11). Among those who
attended one or more groups, median levels of attendance were higher among those randomised
to group art therapy (11 for the art therapy compared with five for activity groups, p = 0.04). In
addition, 44 (31%) of the 140 participants allocated to group art therapy and 30 (21%) of the 140
allocated to activity groups attended 10 or more groups (p = 0.06). The number of participants
per art therapy group ranged from zero to six (mean attendance = 2.4, SD = 1.1). The number of
participants per activity group ranged from zero to nine (mean attendance = 2.1, SD = 0.9).
© Queen’s Printer and Controller of HMSO 2012. This work was produced by Crawford et al. under the terms of a commissioning contract issued by
the Secretary of State for Health.
17