DOI: 10.3390/buildings16204110 ISSN: 2075-5309

Recurrent Supportive Design Themes in Public Project Descriptions of 18 UK Maggie’s Centres

Leyi Sun, Can Zheng, Shuai Ling, Yikang Sun

Public project descriptions record architectural intentions, but they cannot establish how buildings are used or whether they improve health. This study compares documented supportive design themes in a fixed purposive cohort of 18 UK Maggie’s Centres whose reported building or opening years span 2003–2021. Seventeen project descriptions from Maggie’s and one from Steven Holl Architects were captured on 5 October 2026. Ten categories structured the authors’ documentary analysis. Two separate AI passes served as an auxiliary check; the authors verified the coding and results. Agreement between the two AI runs before adjudication was 169/180 (93.9%; Cohen’s κ = 0.878). This is a descriptive measure of AI-run repeatability, not human intercoder reliability. Agreement for identifiable architectural form was weaker (14/18; κ = 0.265). Nature contact appeared in 18 descriptions and identifiable architectural form in 17. Domesticity/kitchen occurred in 11, daylight/outward view in 10, art/symbolic narrative in nine and sensory material character in eight. Hospital/campus transition appeared in seven descriptions, social/private spatial choice and flexibility in five each, and environmental strategy in three. Eligible descriptions contained 146–415 words. Their exploratory length–theme association was positive (Spearman’s ρ = 0.383, p = 0.116), leaving document-length bias unresolved. Source-exclusion and category-merging checks retained the prominence of nature and formal description, subject to the weak FI agreement. Co-documentation indicates overlap in narratives, not a functional or causal design mechanism. The preliminary source-linked framework is intended for comparing architectural project narratives and framing future building and user research. The findings apply only to these 18 descriptions and do not estimate programme-wide prevalence or validate therapeutic outcomes.

More from our Archive