Firsthand owner decisions, practice and household runway, debt and guarantees, delays, unpaid management time, advisor incentives, staffing, payer friction, entry-mode changes, and owner-defined outcomes.
Research infrastructure · protocol v0.3.0-draft
Prepare the owner case series before asking anyone to participate.
DenQAI needs longitudinal owner evidence. It does not need an uncontrolled story form. This wave publishes the draft questions, evidence rules, privacy boundary, consent design, analysis plan, and launch gates while keeping intake closed.
Current statusIntake closedProtocol design and independent governance review
Participants enrolled0No one is being recruited
Files accepted0No upload or submission channel
Published findingsNoneNo owner cases collected or published
The governing question
This may be research. A website cannot determine its own exemption.
The proposed case series is a systematic investigation designed to produce knowledge beyond one owner’s decision, so it may meet the federal definition of research in 45 C.F.R. § 46.102. Interaction with living owners or use of identifiable private information can also raise human-subject questions. Whether the Common Rule applies, whether an exemption is available, and who may make that determination depend on funding, institutional coverage, methods, identifiers, and other law.
Before enrollment, DenQAI will require a dated written determination from a qualified independent reviewer or appropriate IRB/institutional official. The public protocol will not call itself “exempt,” “not human-subjects research,” or “HIPAA compliant” without that documented basis.
Pilot boundary
Study owner mechanisms—not patients, accusations, or market gossip.
Patient-level information, chart images, clinical-necessity adjudication, employee allegations, named-practice rankings, advisor blacklists, malpractice conclusions, or prevalence estimates.
Up to 20 voluntarily enrolled adult U.S. current or former owners in a maximum-variation pilot. The ceiling limits exposure; it is not a statistically representative sample.
A structured baseline and evidence index, then proposed 6- and 12-month follow-up. Any 24- or 48-month extension requires a revised protocol and renewed governance review.
Pre-specified questions
Ask what records can help answer.
The pilot is designed to identify mechanisms and improve evidence requests. It cannot estimate how common a mechanism is among all dentists.
- 01
Which mechanisms create first-48-month practice and household cash strain?
- 02
How do guarantees, delay, capital calls, and unpaid management time change the ownership-versus-employment tradeoff?
- 03
Which advisor incentives or information boundaries changed a decision, and what contemporaneous evidence supports that account?
- 04
Which facts caused an owner to proceed, redesign, delay, renegotiate, sell, close, or walk away?
- 05
Which plausible alternative explanations survive a within-case evidence review?
Evidence treatment
One account can be important without becoming a statistic.
Direct contemporaneous record
A target, transaction, accounting, contract, or operating record tied to the proposition. Its verification state still depends on reconciliation and review.
Reconciled summary
A participant or reviewer summary tied to supporting records and at least one independent control. State which records and totals were checked.
Attributable firsthand account
The participant directly experienced the event. Record what is remembered, what is documented, and what remains unverified.
Inference or unverified report
An interpretation, memory, hearsay report, or unsupported claim. Use it to generate a question, not as a verified finding.
No completeness score.
Each material proposition keeps source kind, support relationship, verification state, period, contradiction, reviewer, limitation, and public-use status separate. Missing evidence cannot be averaged away.
Analysis plan
Preserve chronology, contradictions, and negative cases.
Separate what the participant knew at the time from hindsight, then tie turning points to dated records and decision gates.
Compare mechanisms across cases without treating case frequency as population prevalence.
Record evidence against the preferred explanation and preserve unresolved disagreement.
Actively look for owners who faced the same exposure without the same outcome—or reached the same outcome through a different mechanism.
Use broad periods, ranges, and geography; suppress small cells and unique combinations after a qualified re-identification review.
Report bounded observations such as “observed in X of Y pilot cases,” never causal, national, or failure-rate claims from a purposive pilot.
Participant control
Consent is a process, not a checkbox at the bottom of a form.
Recruitment and any compensation must be neutral and must not depend on the owner’s outcome or viewpoint.
A small community and a distinctive timeline can identify a practice even after names are removed.
Reviewer disagreement, unresolved contradiction, and requested corrections remain visible.
The cutoff and correction path must be approved before enrollment, not invented after a request.
Seven launch gates
Enrollment remains closed until every gate has dated evidence.
A green design note is not approval. The public status must name the reviewer, scope, date, expiration, and unresolved condition without inventing credentials.
Required ownerQualified independent human-subjects or research-governance reviewer
Evidence to clearDated written determination addressing 45 C.F.R. Part 46 applicability, human-subject status, exemption or review path, and other applicable law.
Required ownerResearch lead plus independent methods reviewer
Evidence to clearVersioned questions, eligibility, recruitment ceiling, data dictionary, independent evidence axes, negative-case method, and publication rules.
Required ownerResearch-governance and privacy reviewers
Evidence to clearApproved plain-language consent, voluntariness, compensation, risks, data use, participant review, withdrawal timing, and contact route.
Required ownerQualified privacy counsel and de-identification reviewer
Evidence to clearData-flow map, prohibited fields, code-key separation, release transformation, small-cell suppression, re-identification test, and state-law review.
Required ownerSecurity lead and independent assessor
Evidence to clearApproved collection channel, least privilege, encryption, logging, backup, retention, deletion, incident response, vendor review, and tested recovery.
Required ownerResearch-governance reviewer
Evidence to clearNeutral recruitment text, inclusion/exclusion rules, fixed compensation if any, no outcome-contingent payment, conflict disclosure, and anti-retaliation safeguards.
Required ownerMethods editor and participant-protection reviewer
Evidence to clearParticipant fact-check, attribution rule, small-cell rule, withdrawal cutoff, correction path, adverse-event escalation, and non-prevalence language.
Privacy architecture
Do not collect what the pilot does not need.
The draft design prohibits patient information and separates any future identity/contact key from the coded case file. Public analysis would use broad ranges and periods, suppress rare combinations, and require qualified re-identification review. Removing names alone is not de-identification.
HIPAA uses specific de-identification paths—Expert Determination or Safe Harbor—for protected health information held by covered entities. DenQAI’s no-PHI rule is stricter for this pilot: owners should not submit patient-level material even if they believe it has been de-identified.
Controlled work products
Inspect the draft before any intake system exists.
Pilot protocol draftPurpose, boundary, questions, recruitment ceiling, evidence method, analysis, governance, and launch rules.Markdown · draft v0.3.0-draft ↓Case data dictionaryField-level purpose, allowed values, sensitivity, required status, public-release treatment, and source rule.CSV · controlled dictionary ↓Case evidence indexRecord category, period, source control, source kind, support relationship, verification state, contradiction, reviewer, limitation, and release status.CSV · fictional example row ↓Consent design previewNot a usable consent form: the elements that must be approved before recruitment.Markdown · intake closed ↓
Local preparation only
Structure a case without sending it to DenQAI.
The preparation tool runs in the active browser tab, rejects declared privacy hazards, and exports a draft file to the user’s device. It is not enrollment, consent, submission, or approval.
Open the local case builder