Result-blind framework development and prospective evaluation of treatment-strategy identifiability after response/control in atopic dermatitis (operative prospective registration)
Year: 2026
Type: preprint
Abstract: This is the operative prospective registration for a result-blind methodological evaluation of whether independent reviewers can reproducibly determine whether newly retrieved human atopic dermatitis/eczema study records identify MAINT_VS_WITHDRAW after a prespecified response/control state. The target strategy comparison is continuation of the same effective treatment versus direct complete withdrawal at the same post-response/control time zero. Tapering, dose reduction, interval extension, switching, delayed initiation, and uncontrolled discontinuation are not silently treated as direct withdrawal. The primary analysis unit is the arm/strategy path within a linked study/cohort. Primary methodological outcomes are rule determinability/classifiability and pre-adjudication reviewer agreement; secondary outcomes include adjudication, lineage resolution, unresolved classification, erroneous target-layer inclusion prevention using a locked design-only challenge set, strategy diversity, missing design domains, and protocol deviations/amendments. No clinical outcomes, effect estimates, real instrumental variables, harmonization, linkage disequilibrium, Mendelian randomization, pooling, or meta-analysis will be performed in this registration. An earlier OSF registration, h3wsu, was registered on August 16, 2026 at 00:19:04 and is under embargo. On August 24, 2026, its signed-in OSF Files inventory was verified to contain no files. The present registration is therefore a new operative registration with the reviewed method attachments listed in the local candidate manifest; it is not an amendment or retroactive attachment to h3wsu. No formal prospective retrieval, screening, coding, lineage resolution, adjudication, or outcome-bearing analysis was conducted before this registration. The historical development stage was unregistered, retrospective, and result-blind; it is disclosed as development context and is excluded from prospective denominators. The attached method package fixes the population scope, source/platform matrix, reviewer coding, protocol freeze, and result-blind boundaries for prospective execution. The five method attachments are protocol_v1.1.md, analysis_plan_v1.1.md, source_platform_matrix_v1.1.csv, exact_search_strategies_v1.1.md, and eligibility_adjudication_codebook_v1.1.md.
Author: Haoyang Mu
Language: English
Subfield: Statistics, Probability and Uncertainty
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7204587751