Designing a Quality Assurance Framework for Evaluating Administrative Reliability in Home Based Healthcare Transitions

Abstract

A companion research program has argued that administrative reliability, not a single technology or score, governs continuity of home health care, and has proposed a standardized protocol suite to act on that governance diagnosis. Neither contribution can be tested credibly without a third: a measurement framework capable of establishing that a given administrative reliability indicator is reliable, valid, and comparable across the agencies, patients, and time periods to which it will be applied. This paper builds that framework. It grounds administrative reliability measurement in the Donabedian structure-process-outcome tradition and in classical test validity theory, and proposes a quality assurance architecture with four indicator classes, structural, process-reliability, criterion, and equity-differential, each carrying an explicit psychometric requirement rather than an assumed one. The paper specifies nine validation propositions covering interrater reliability, internal consistency, criterion validity against Home Health Value-Based Purchasing performance, and measurement invariance across rurality and language subgroups, and lays out a four-stage validation plan, content review, reliability piloting, criterion and construct validation, and equity-differential testing, before any indicator is certified for routine agency use. The paper is explicit about what psychometric validation can and cannot establish in a regulated, accredited industry: validated measurement properties support comparability and fair inference, not automatic causal or policy conclusions, and a framework that blurred that distinction would misrepresent its own evidentiary status. No indicator proposed here has been validated empirically; the contribution is the validation architecture itself, offered as the measurement layer this research program's governance diagnosis and protocol design require before either can be tested at scale.