Arts for Wellness (AFW): Arts-Based Social Prescribing During Inpatient Rehabilitation for Older Adults in Singapore

Background: Social prescribing enables health professionals to refer patients to non-clinical, community-based support. Arts-based social prescribing has been associated with improved wellbeing, but most evidence comes from uncontrolled community studies in Western settings. No randomised trial had evaluated arts-based social prescribing introduced during inpatient rehabilitation, and none had been done in Asia.

Design: Parallel-group, open-label, individually randomised controlled trial with a nested qualitative strand (QUAN[qual]) at Outram Community Hospital (OCH), a step-down facility providing subacute rehabilitation, where social prescribing delivered by Wellbeing Coordinators has been routine care since 2019. Participants were randomly assigned 2:1 to AFW plus standard care or to standard care alone using a computer-generated permuted-block sequence prepared by a biostatistician with no role in recruitment or outcome assessment. Allocation was concealed until enrolment was complete. Participants, artists, and research coordinators were not masked; outcomes were self-reported.

Original design and implementation: The original protocol and sample-size calculation treated dance, literary arts, and control as three groups in a 1:1:1 allocation, equivalent to an overall 2:1 allocation to an AFW intervention versus control. During implementation, participants were randomised directly 2:1 to AFW plus standard care or to standard care alone; thus, the overall intervention-to-control allocation ratio was preserved. Within the AFW arm, however, participants were not randomised between dance and literary arts. The art form was determined by which programme was scheduled on the ward during each participant's admission week. Accordingly, the comparison is AFW (dance + literary arts) versus standard care..

Intervention: AFW comprised structured group sessions delivered by National Arts Council artists in one of two art forms. The dance programme ("Move to the Rhythm of Life") drew on Malay dance principles and used culturally familiar objects (batik cloths, sarongs, fans, drums). The literary arts programme used modular storytelling prompts, reminiscence, character-building, and group expression. Sessions lasted 45-60 minutes, twice weekly for 3 weeks (six sessions), with artists adapting content in real time to participants' language, mobility, energy, and confidence. Programme design drew on the PERMA wellbeing model and the INNATE framework for active ingredients of arts-in-health activities. After discharge, participants were offered two virtual Social Hour sessions (Zoom) within 8 weeks of their final session, with volunteer support for those less confident with technology, a Take-Home Kit specific to their art form, and linkage to community arts programmes (e.g. at Active Ageing Centres). The control group received standard community hospital care, including routine social prescribing.

Outcomes: The primary outcome was mental wellbeing (14-item WEMWBS) at T1, 4 weeks after baseline. Secondary outcomes were WEMWBS at T2 (8 weeks after T1) and T3 (4 months after T1); symptoms of depression (PHQ-9) and anxiety (GAD-7); health-related quality of life (EQ-5D-5L utility index using the Singapore value set, and EQ VAS); patient satisfaction (study-specific patient-reported experience measure, intervention group, T1); and health-care utilisation (length of stay and 30-day readmission). Attendance at AFW sessions and Social Hour was recorded as a process measure. Questionnaires were administered at T0 (baseline), T1, T2, and T3, each within a window of 7 days either side.

Analysis: Intention to treat. The primary outcome was analysed with a constrained longitudinal data analysis model including time, treatment-by-time interaction, and an unstructured covariance matrix, adjusted for age, sex, and ethnicity, with missing data assumed missing at random. Sensitivity analyses were also performed: unconstrained longitudinal model; ANCOVA for the T1 effect under multiple imputation, complete-case, and baseline-observation-carried-forward assumptions. Secondary outcomes were analysed similarly without adjustment for multiplicity.

Qualitative strand: Researchers independent of programme delivery conducted semi-structured interviews with AFW participants purposively sampled from those who had attended at least four sessions (to capture variation in art form, mobility, language, and engagement), and focus groups with dance artists, literary arts artists, health-care professionals (Wellbeing Coordinators, physiotherapist, doctor), and programme implementers. Interviews and focus groups were audio-recorded, transcribed verbatim, and translated into English if conducted in Mandarin, other Chinese dialects, or Malay. Data were analysed using reflexive thematic analysis, with themes subsequently read against the PERMA model and the Practical, Robust Implementation and Sustainability Model (PRISM). The two strands were analysed independently and integrated at interpretation using a joint display.

Reporting follows CONSORT 2025, GRAMMS, and COREQ.