This paper reconsiders long-term disaster recovery assessment by distinguishing declining psychological symptoms, visible restoration of housing and services, and the recovery of everyday roles, relationships, routines, place relations, and future possibilities through which life remains meaningful.
A structured conceptual narrative review integrates disaster mental health research, critical disaster studies, community and place-based recovery scholarship, well-being theory, and long-term Japanese disaster research. A bounded supplementary verification search examined 38 publications and retained 24 to assess the framework's breadth across concepts, geographical settings, and disaster contexts.
The review defines the recovery paradox through four linked features: dimensional divergence, temporal asynchrony, interpretive substitution, and consequences for recovery management. It proposes a dynamic layered framework in which symptom, visible restoration, and lifeworld evidence are interpreted together rather than treated as interchangeable indicators of recovery.
Monitoring and support decisions should match different forms of evidence to clinical, material, welfare, relational, and role-based interventions. Support may need to continue, be redirected, be reduced in stages, or end according to the combined pattern across the three assessment levels.
Recovery policies should avoid treating symptom improvement or visible reconstruction as sufficient evidence of recovery. Incorporating lifeworld conditions can help reduce premature withdrawal of support and make long-term recovery decisions more responsive to social and relational needs.
The paper moves beyond observing that recovery dimensions differ. It specifies when evidence from one dimension is overgeneralised into a judgement about recovery or non-recovery as a whole and how that misreading can distort support continuation, redirection, reduction, or closure.
