Design an EHR migration reconciliation sample covering record populations, risk, selection, expected results, reviewer evidence, and exception follow-up.
Migration reconciliation should combine completeness checks with known record-level cases that test identity, relationships, historical meaning, and actual workflow usability. A matching total is useful evidence, but it does not answer every acceptance question.
This guide describes an operational review design for behavioral health migrations. It is not a statistical sampling protocol or a universal acceptance standard. Define the actual scope and seek qualified input when the project needs formal inference, legal review, or specialized financial validation.
EHR migration reconciliation sample design starts with questions
List what the organization needs to establish: completeness, correct identity, episode relationships, readable content, historical dates, signature context, future scheduling, opening balances, or another scoped question.
The ASTP/ONC EHI export test method provides a certification-related export reference. The receiving migration still needs its own evidence for mapping, representation, and usability. An export and an accepted destination workflow are different outcomes.
Assign the review owner for each question. Technical staff can compare files and counts, while authorized clinical, scheduling, identity, and finance reviewers evaluate meaning and operational use.
Establish source populations and totals
Record the source system, extract version, cutoff, categories, date coverage, filters, and exclusions. A destination total cannot be interpreted without knowing the source population it is expected to represent.
Compare useful category totals and identifiers where appropriate. Patient, episode, encounter, document, attachment, appointment, and transaction counts are different units. Do not combine them into one migration-complete percentage without explaining the basis.
Preserve expected differences. Some categories may intentionally use a different destination representation under an approved decision. The reconciliation should distinguish that limitation from an omission or mapping defect.
Select known ordinary and exception cases
Use fictional or appropriately authorized records with known expected relationships. Include ordinary examples and cases likely to reveal incorrect assumptions. Label targeted selection accurately rather than calling it random or representative without a basis.
The ASTP/ONC SAFER patient identification guide provides a primary safety reference. Include identity and association questions rather than assuming matched demographics establish that every linked record is correct.
For a fictional cohort, use same-name people, changed demographics, several episodes, a signed note with an amendment, a multi-page attachment, a recurring appointment exception, and an open financial item. Tailor the cases to the actual scope.
Test relationships, not only individual fields
A correct field can belong to the wrong person, episode, encounter, or claim. Trace the links that staff use to understand the record. A document title and date alone may not reveal an incorrect association.
Ask reviewers to explain the destination history without source-system coaching. They should identify the relevant episode, authorship, dates, versions, and next action. If the explanation depends on a migration specialist's memory, the representation needs clarification.
Use totals to identify possible gaps, then investigate records. One missing item and one duplicate can offset each other. Record-level validation can detect that problem even when aggregate counts agree.
Include actual receiving workflows
Have ordinary users locate, review, schedule, export, or continue work as relevant to their roles. Data can be present but unusable because it lands in an unexpected field, inaccessible view, or disconnected attachment list.
| Review layer | Evidence question |
|---|---|
| Completeness | Do approved source categories and identifiers reconcile? |
| Identity | Does information belong to the correct person? |
| Relationships | Are episodes, encounters, documents, and transactions linked correctly? |
| Meaning | Are dates, authorship, versions, and source context understandable? |
| Usability | Can ordinary staff perform the intended receiving task? |
| Exceptions | Are uncertain or defective items visible and owned? |
| Final changes | Does cutover reflect updates after rehearsal? |
Keep detailed evidence appropriately restricted. A general project report can identify case references and outcomes without containing full patient or financial records.
Record discrepancies with specific owners
Classify differences as source issue, mapping issue, missing record, wrong association, representation limitation, workflow problem, or unresolved question. A generic migration defect label can conceal the person able to resolve it.
Include the source reference, observed result, expected result, evidence, owner, next action, and decision. Keep observation separate from interpretation so reviewers can discuss disagreements without losing the original facts.
Do not change expected results merely because the imported output looks convenient. If the organization approves a different representation, record the revised scope and acceptance basis explicitly.
Retest affected scope after changes
A mapping correction can affect more than the record that exposed it. Identify the categories and relationships influenced by the change and rerun appropriate tests. A fixed example does not establish that every related record is now correct.
Preserve extract, mapping, import, and validation versions. The organization should know which evidence applies to the final rehearsal and which applies to cutover. A passing earlier run does not automatically cover later source or configuration changes.
Use a known late change after the rehearsal extract to test the final-update process. Confirm it arrives once, retains relationships, and does not overwrite approved corrections unexpectedly.
Make acceptance a documented decision
Summarize passed checks, unresolved issues, approved limitations, manual controls, and blockers. Have authorized owners accept their respective scope. The technical importer should not sign off on all clinical and financial meaning alone.
If the project needs a formal statistical or audit methodology, obtain appropriate specialist design. This practical checklist does not establish confidence levels or guarantee correctness across every record from a small targeted sample.
Put acceptance evidence and responsibilities into the written vendor scope. Distinguish available migration capability, configuration, source dependencies, manual review, and unverified claims. A broad everything transfers statement is difficult to evaluate without those details.
Keep the evidence useful after launch
Retain the final reconciliation and unresolved-work ownership in an authorized location. Staff may discover a historical question later and need to know the approved representation or source route.
Review post-launch findings against the acceptance cases. Use recurring patterns to improve mapping, training, or source governance rather than treating every issue as an isolated surprise. Preserve the actual correction history.
Ease Health publishes this guide and offers a unified CRM, EHR, and RCM platform for behavioral health. Ask vendors to demonstrate the proposed reconciliation with known cases and ordinary users. Related material includes the EHR overview, migration identity guide, and historical notes validation guide.
The useful result is a review that explains what was checked, what passed, and what remains uncertain. That is a stronger basis for acceptance than a matching total without record-level meaning.
Related reading
Frequently Asked Questions
Does a matching record count prove a migration is correct?
No. Counts can match despite wrong associations, missing content, duplicate offsets, or unusable history. Combine totals with record-level and workflow validation.
Is there one universal migration sample size?
No sample size is prescribed here. Define the purpose, categories, risk, selection method, and acceptance scope. Obtain appropriate specialist input if statistical inference or a formal methodology is needed.
What should exception cases cover?
Include same-name people, multiple episodes, signed and amended records, attachments, recurring appointments, financial corrections, late changes, and unresolved mappings relevant to the actual scope.
Who signs off on migration reconciliation?
Assign authorized owners for clinical, identity, scheduling, financial, technical, and privacy decisions as relevant. Technical job completion does not establish every department's acceptance.
