Ten sessions of support against one session of crisis planning
- The study and its design. The Structured Psychological Support (SPS) study was a multicentre, two-arm, researcher-masked, randomised controlled superiority trial run across seven NHS Trusts in England, published in The Lancet Psychiatry in 2026 by Crawford and colleagues. Recruitment ran from 7 February 2023 to 31 January 2024 and the last follow-up assessment was completed on 28 February 2025. The trial was registered as ISRCTN13918289 before data collection, the statistical analysis plan was published before the database was locked, and the work was funded by the National Institute for Health and Care Research. The authors declare no competing interests.
- Who entered, and on what threshold. Entry required age 18 or over and a score of 4 or more on the Standardised Assessment of Personality Abbreviated Scale, a threshold for probable personality disorder rather than a named diagnosis. The authors tie that choice to the change in ICD-11, which classifies personality disorder by severity rather than by type. Exclusions were a co-existing organic or psychotic disorder, current or imminent psychological treatment for personality disorder, and enrolment in another study. 336 adults were randomised, 180 to SPS plus treatment as usual and 156 to enhanced treatment as usual, in a 1.15:1 ratio. 251 (75%) were women, mean age 34.8 years (SD 13.2, range 18 to 68), and 281 (84%) were White. Participants had first reached mental health services a mean of 13.5 years earlier (SD 10.4). 247 participants, or 74%, met SCID-II criteria for borderline personality disorder, a figure the authors report and then test separately.
- The two procedures being compared. SPS is up to ten individual sessions of 45 to 50 minutes, usually fortnightly, in person or online, with a review session offered afterwards. The first two sessions assess how the patient understands their difficulties and how they cope; the pair then agree a psychological focus and one or more skills. The comparator, enhanced treatment as usual, was a single mental health crisis planning session, generally delivered by a psychiatrist, alongside existing primary or secondary care. SPS cost £878 per participant, the crisis planning session £86.
- The primary outcome. Social functioning on the Work and Social Adjustment Scale at 12 months, analysed by intention to treat, showed no difference between arms: standardised coefficient 0.12, 95% CI -2.14 to 2.38, p = 0.92, Cohen's d 0.10. The subgroup restricted to participants meeting borderline personality disorder criteria gave a coefficient of 0.92, 95% CI -0.92 to 2.76, p = 0.33. The per-protocol analysis gave 2.77, p = 0.080. Follow-up at 12 months was completed by 152 (84%) in the SPS arm and 132 (85%) in the control arm.
- Secondary outcomes, delivery and cost. Two secondary measures favoured SPS by small margins: emotional dysregulation (DERS-16) at 4.29, 95% CI 0.96 to 7.63, p = 0.012, d 0.19; and self-rated global improvement (CGI) at 0.70, 95% CI 0.11 to 1.29, p = 0.020, d 0.25. Depression, anxiety, personality dysfunction, self-harm and suicide attempts showed no difference. Participants received a median of 7 sessions (IQR 3 to 10) and 21 of 180 (12%) attended none. The probability that SPS was cost-effective was 0.34 to 0.39 at the £20,000 to £30,000 per QALY threshold. Two participants died during follow-up, both in the SPS arm; no serious adverse event was judged related to study participation.
The opening two sessions
The practitioner's first move is to hand the patient a written account of what the coming weeks will contain. After that, the opening two sessions go on one task: working out how this person understands their own difficulties, and what they already do to cope with them. Only when those two sessions are done do practitioner and patient settle on a psychological focus and choose one or more skills to work through, among them mindfulness, interpersonal effectiveness and mentalising.
Sessions last 45 to 50 minutes. Most ran fortnightly, some weekly at the participant's request, and 23% were delivered online. Ten is the ceiling, and a review session is offered once the planned ones are finished.
One session of crisis planning
The other arm of the trial was not an empty one, and this is the part that decides how the headline result reads. Everyone allocated to enhanced treatment as usual was offered a single mental health crisis planning session, generally delivered by a psychiatrist, on top of the primary or secondary care they were already receiving. That session cost £86 a head. The ten-session package cost £878. Of the 129 control participants who reached the offer, 85 took an updated crisis plan and 44 turned the meeting down.
The WSAS result
The primary outcome was the Work and Social Adjustment Scale at twelve months, analysed across all 336 randomised adults. The arms came apart by a standardised coefficient of 0.12, 95% CI -2.14 to 2.38, p = 0.92.
Narrowing to the 247 participants who met SCID-II criteria for borderline personality disorder – 74% of the sample, and a number the authors put forward themselves – returned a coefficient of 0.92, p = 0.33. Two secondary outcomes did move in favour of the longer arm, emotional dysregulation at 4.29 (p = 0.012) and self-rated global improvement at 0.70 (p = 0.020), both under an effect size of 0.25. Depression, anxiety, personality dysfunction, self-harm and suicide attempts held level.
Practitioners, dose and cost
Seventy-six staff delivered SPS, mostly psychologists (26) and nurses (21), with a median of six years in the NHS behind them. Each had nine hours of training across three sessions plus fortnightly group supervision, attended 76% of the time.
The protocol assumed every practitioner would treat four or five participants. Turnover meant more than 70% treated one or two. Participants received a median of seven sessions out of a possible ten, and 21 of 180 attended none at all. The per-protocol analysis, restricted to those who did receive the sessions, returned the same answer as the main one (p = 0.080), as did the analysis dropping the therapist random effect (p = 0.39).
The comparator was not nothing. It was one crisis planning session at £86 a head, set against £878 for ten sessions of support.
This trial tested one brief individual intervention against one crisis planning session on one primary outcome over one year, and its result should not be read past those boundaries. The authors name several constraints on delivery. There was no high-quality information on treatment fidelity, because audio recordings of sessions could not be collected or analysed; fidelity was explored qualitatively and through practitioner pro formas instead. Each practitioner treated far fewer participants than planned, with more than 70% treating one or two rather than the four or five the protocol assumed, so practitioners may not have built the experience and confidence the intervention needs. Training was limited to nine hours across three sessions. Participants had been in contact with mental health services for a mean of 13.5 years before entering, and the authors suggest SPS might do more if offered earlier in that contact. Practitioners reported taking longer to build a therapeutic relationship than expected, leaving little time to rehearse skills within the median of seven sessions delivered. The small number of participants who were not women restricts generalisability. The trial recruited by a screening threshold rather than by diagnosis, yet 74% still met criteria for borderline personality disorder, so it does not speak separately to other personality disorder diagnoses. A null result on this outcome, in this sample, over this period is not evidence that brief interventions in general are without effect, and the authors themselves call for wider access to longer evidence-based programmes rather than for less provision.