Abstract
Rehabilitation in long-term conditions requires not only clinical capacity but also sustained patient engagement, yet retention is commonly reduced to crude completion metrics. This theory-informed conceptual mini-review used a purposive, non-systematic synthesis of literature identified from the authors' existing knowledge base, targeted PubMed/MEDLINE searches last updated on 9 August 2026, and backward reference checking; the final article cites 76 sources. Retention is reframed as six related phases comprising referral and eligibility, uptake or initiation, early attendance, ongoing attendance or dose, completion or planned discharge, and post-program maintenance, while time to dropout is treated as a longitudinal metric. Across these phases, disengagement may reflect interactions among clinical severity, symptom interpretation, illness beliefs, distress, motivational climate, therapeutic relationships, social context, and structural access. Four author-generated hypotheses concern psychological flexibility and acceptance, illness coherence and rehabilitation self-efficacy, autonomy-supportive care and therapeutic alliance, and social support and peer belonging. Direct evidence is concentrated in selected rehabilitation settings, whereas several stage-specific and cross-condition pathways remain theoretical extrapolations. Rehabilitation resilience is therefore proposed as a provisional interaction between patient adaptive resources and responsive care systems, not as a validated trait or a synonym for observable retention. Structural barriers may prevent participation directly, and discontinuation may be appropriate when treatment is unsafe, ineffective, burdensome, or inconsistent with informed patient goals. The framework is intended to guide prospective, condition-specific testing rather than imply causal validation or scientific consensus.