Queries

A query is a managed question about data: a discrepancy with the source, a clarification, a remark. Queries are the primary data-cleaning tool. They look and behave the same everywhere: in the form, in the study worklist, and in your personal task worklist.

0:000:00

Queries console: auto, manual and medical queries for the whole study in one worklist, with filters and aging

Query types

Type	Who opens it	When
Auto	System	An edit check with a query action fired on save
Manual	Monitor, data manager, PI	A remark on a value that a rule cannot check
System	System	A process inconsistency (e.g., a conflict after randomization)
Medical	Medical monitor	A clinical question about safety data

Query lifecycle

1. Open
   • cancelCancelled
   answer
2. Answered
   • rejectOpen · new round
   accept
3. Closed
   • reopenReopened

The query's main path plus the additional transitions.

Status	Meaning
Open	Awaiting a response
Answered	A response has been given; awaiting review and closure
Closed	Discrepancy resolved
Reopened	Returned to work after closure
Cancelled	Deemed opened in error (terminal)

The system closes auto-queries itself

An automatic query (the “Auto” type) is driven by the rules engine: it opens when an edit check with a query action fires on save, and closes itself as soon as the data is corrected and the check no longer fires (reconciliation). An auto-query cannot be reopened by hand — it “reopens” through a new violation of the same rule.

Open a query

A manual query is opened right in the form — at two levels:

• On a field — from the field menu, choose “Open query…”. A field query is available once the item has a value (while the field is empty the item hints: “available once the item has data”).
• On the form — the “Query on the form…” button at the top of the form. Use it for a remark not tied to one filled field: a missing (empty) value, a cross-field inconsistency, or a question about the whole form.

The system opens an automatic query itself when an edit check with a query action fires on saving a value.

Do not quote blinded values

If a field is blinded, the system will warn you: do not include the value in the query text — query text is not masked. Describe the problem without revealing the value itself.

Answer

The response usually comes from the site coordinator. They either explain the value in text or correct the data (the edit goes through a reason for change) — and the query moves Open →

Answered. After the answer, the query awaits closure by a monitor or data manager.

Close, reopen, cancel

• Close — accept the answer: Answered → Closed. Rejecting the answer returns the query to work as a new round.
• Reopen — return a closed query to work with a reason.
• Cancel — mark a query as opened in error (with a reason). Cancellation is terminal: a cancelled query cannot be reopened.

Closing is not cancelling

Closing means the discrepancy is resolved; cancelling means the query should not have been opened. For data-quality metrics the two are kept fundamentally distinct.

Notifications are addressed, not broadcast

A status change notifies exactly the people affected, not “everyone”: an answer → whoever opened the query; a reopen → the author of the last answer; a close → both. Opening a query broadcasts to no one — the new query simply appears in the “Tasks” worklist of those who hold the permission to work it within its scope (see below). The notification carries only a label and a link; the clinical value never travels in it.

The discussion thread

All exchanges on a query live in a single thread: comments, answers, and resolutions. The thread counts rounds (how many times the query went back for rework) and reopens. A single thread preserves the discussion history and audit — instead of spawning new queries for the same problem.

Worklist and bulk actions

It is convenient to work through queries not one casebook at a time but in a worklist. It exists at three levels:

• the study worklist — all queries (the Queries tab);
• the subject worklist — the Queries tab in the casebook;
• “My Worklist” (Tasks) — queries that await you specifically.

The worklist is built from your permissions

Your worklist contains exactly those queries where you have the right to answer or close within their scope: when access changes, worklists rebuild themselves. Assignment (below) is a label ON TOP of this rule, not a replacement for it.

The worklist supports filters (status, type, site, form, age, “assigned to me”) and bulk actions:

1. Check several queries with checkboxes.

2. Choose a bulk action — Close, Reopen, or Cancel.

3. The action is applied to each selected query individually and audited separately.

A bulk action is not “all or nothing.” If the action is not allowed for some queries, they land in “skipped” with a reason: “no permission”, “scope under hard lock”, “invalid state transition”, “not found”. In the end you will see “applied N, skipped M”. A worklist row leads to the form with focus on the field — editing always happens in its original context.

Assigning a query

A live query can have an assignee — the person who took it (or was asked) to see it through. The query thread shows an “Assignee” row with “Assign…” (a pick from candidates), “Take it” and “Unassign” actions; assignment is managed by roles holding the query.assign permission (the data manager and the monitor by default). The assignee receives a notification, the query rises to the top of their “Tasks” with an “assigned to me” label, and worklists gain an assignee column and filter.

Assignment is a label, not a permission

Only someone who can already work the query in its scope (holds the permission to answer or close) can be assigned — the system computes candidates from grants. Assignment adds and removes no permissions and does not hide the query from others: the worklist is still built from permissions; the label merely records who owns it.

Deadlines and aging (SLA)

Each query shows its age — how many days it has been open. The deadline threshold (SLA) is configured at the study or site level; worklists highlight overdue queries, and reports compute aging by age buckets (0–3, 4–7, 8–14, 15+ days).

The system does not close queries on timeout

SLA is a metric and a highlight, not a status transition. An overdue query stays open until a human resolves it. The system never “closes on time.”

Source Data VerificationVerifying data against the source; a discrepancy → a manual query.

Medical reviewMedical queries and review of safety data.

Reason for change (RfC)How an answer with a data edit records the reason for change.

Status lifecyclesThe full query state machine and the other status axes.