Skip to main content

Preventing Duplicate Check-Ins: Handle Double Clicks, Retries and Two Desks

Prevent duplicate check-ins by recognising the intended visit across repeated clicks, uncertain retries and requests from different desks. A short button delay can reduce accidental clicks, but the server also needs appropriate operation and visit identity. Preserve legitimate return visits and different family members; do not use one phone number or one patient-per-day rule as a universal duplicate test.

By Bractis · Updated · 9 min read

Indian reception teams and clinic owners evaluating queue software during busy walk-in sessions, slow connections and work shared across multiple desks.

A second click is not always a second visit

A check-in creates or advances a particular clinic visit. Before designing duplicate protection, define the patient, appointment or walk-in intent, service, location and current encounter state that identify that visit.

Two requests can refer to the same intended visit even if they arrive seconds or minutes apart. A patient may still be standing at reception after the first response was lost, prompting staff to try again.

The reverse is also true: the same patient may legitimately have more than one episode under the clinic's actual workflow. Returning after an earlier encounter, receiving a different service or continuing a prior visit need explicit distinctions; a blanket daily block can be wrong.

A shared family phone is especially unsuitable as a unique visit identity. Two relatives can arrive together and need separate records and queue entries. Duplicate protection should not merge them merely because reception found both through one contact number.

Keep appointment identity and queue token identity understandable. A token printed twice can still refer to one visit, while two different tokens accidentally created for the same visit represent a reconciliation problem.

Local feedback and server controls solve different parts of the problem

A disabled button and visible in-progress state can stop an immediate accidental repeat on one screen. They cannot prevent another tab, a second computer or a retry after the screen has reloaded from sending a new request.

An arbitrary 600 ms pause is therefore not proof of duplicate prevention. Network responses can take longer, and a duplicate can arrive after any chosen delay. The interface needs an honest operation state rather than a timer presented as a guarantee.

For the same intended operation, a server can use a stable request identity to recognise retries. Stripe's documentation provides a provider example of idempotent requests, with behaviour that differs by API version. It illustrates the principle, not a universal clinic implementation or retention period.

A retry identity should correspond to the same action and relevant input. If the patient, service or appointment changes, the system must not silently treat the changed request as an identical retry. Conversely, generating a fresh identity for every repeat defeats recognition.

Two desks may generate different request identities for the same appointment. The application also needs an appropriate visit-level rule or an explicit conflict process. Request-level replay alone does not necessarily catch independently initiated duplicates.

Apply the control where the record is created or advanced, with appropriate concurrent handling. A visual search showing no active visit can become stale before another desk creates one.

On a narrow screen, scroll the table horizontally to read all columns.

Duplicate-check-in controls and the cases they need to handle
SituationUseful protectionWhat must remain possible
Immediate second clickIn-progress feedback and repeat suppressionClear recovery if the original operation is unresolved
Lost response then retryRecognition of the same intended operationRetrieval of the already-created visit
Two desks, one appointmentVisit-level identity and concurrent handlingA clear existing-visit result for the second desk
Same person, legitimate new episodeExplicit episode/service distinctionA valid new visit under the clinic's actual process
Two relatives, shared phoneSeparate verified patient identitiesIndependent queue entries for each person
Duplicate discovered laterAttributable reconciliation of related recordsPreservation of clinical and financial work already attached

Source: Stripe: Idempotent requests

Look for the existing visit before starting another one

When the response is lost, retain the patient context and any operation reference. Tell staff that the result is not yet confirmed; do not present the missing response as proof that check-in failed.

Look up the current appointment or intended visit through the appropriate authorised route. If the visit exists, return its identity and current queue state. Staff should not need to create another token just to obtain a visible success message.

If no visit is found, follow the configured recovery process, accounting for the possibility of a request still being processed. The product needs a clear rule for verification and retry; a receptionist should not invent one from repeated refreshing.

A patient arriving at another desk should be matched to the correct person and intended episode before any new action. Seeing an active visit may resolve the request, but its presence does not automatically mean the patient has already been seen or should be removed from the queue.

If a duplicate already has work attached, do not simply delete it. Observations, notes, orders or collections may need professional and financial reconciliation. Preserve the relationship and an attributable correction rather than losing evidence.

Treat a wrongly selected family member as an identity error, not a harmless duplicate. The response may require correction of affected downstream records, not just removal of one queue row.

Use a denominator that reflects actual arrivals

Consider an original fictional session with eight patients and nine legitimate intended visits. One patient has an explicitly accepted new episode after an earlier visit; that distinction is part of the example, not a rule that every return creates a new charge or encounter.

There are twelve interface attempts: nine initial attempts and three extra interactions. One extra click is stopped locally, one retry reaches the server with the same operation identity and one second-desk request is recognised as referring to an existing visit.

Eleven requests reach the server because one of the twelve interface attempts was stopped before submission. At the reconciled cutoff, nine visits exist for eight people. Two of the eleven server requests are recognised repeats rather than new visits.

The counts reconcile in two ways: 9 + 3 = 12 interface attempts, and 9 + 2 = 11 server requests. Neither twelve nor eleven is the number of patients seen, and nine check-ins do not prove nine completed consultations.

If a dashboard counted each successful response as a new arrival, a replay that correctly returned an existing visit could inflate activity. Reports should count the actual visit events appropriate to the question.

Keep valid repeat episodes visible while monitoring duplicate attempts separately. A lower duplicate count achieved by blocking legitimate care is not a successful control.

Rehearse concurrency and recovery with the actual desk arrangement

Use synthetic patients and two authorised desk accounts. Try an immediate double click, a delayed response, refresh after submission and simultaneous check-in of the same appointment. Compare the final stored visits, not only the messages each screen displays.

Include two relatives with a shared phone, a valid new episode for the same person and a continuation that should remain within an existing encounter. These cases challenge overly broad duplicate rules.

Repeat one request with changed meaningful details. The system should not silently return a result for the wrong patient or service because a retry identifier was reused incorrectly.

Inspect the queue after recovery. Token order, appointment link, current clinical state and staff attribution should remain understandable. The test is incomplete if it checks only that one row exists but ignores which patient or episode it represents.

Document the retention and recovery limits that the provider actually supports. A retry mechanism may have a bounded lifetime; do not assume that an old request can be replayed indefinitely or that an expired key proves no earlier visit exists.

Create a clear operational response for duplicates discovered after consultation work begins. Prevention reduces the problem, while an accountable correction route handles the cases it does not prevent.

Twelve attempts can correctly end with nine visits for eight people

The fictional session has nine intended visits, one locally stopped extra click and two server-recognised repeats. Twelve interface attempts become eleven server requests and nine stored visits. Keeping those units separate prevents retries from inflating arrival counts or blocking a legitimate new episode.

Reconcile repeated check-in attempts

Use these prompts in your clinic’s own notes. Record the result, responsible person and next action for each item.

Define visit identity
Distinguish a repeated request for one episode from a legitimate new visit and from another family member.
Exercise two desks
Attempt the same synthetic appointment concurrently and inspect the stored visit and both users' outcomes.
Lose a response
Recover an unknown check-in result without blindly creating another token.
Reconcile the counts
Count people, intended visits, interface attempts and server requests separately after recovery.

Questions doctors ask

Does a 600 ms button delay prevent duplicate check-ins?

It may reduce immediate accidental clicks on one screen. It cannot by itself handle delayed retries, another device or two desks acting on the same visit.

Can a shared mobile number be used to block duplicate visits?

Not as a universal identity rule. Family members may share a phone and still require separate patient records and visits.

Should the same patient be blocked from a second visit on the same day?

Not automatically. Determine whether the request is a continuation, duplicate or legitimate new episode under the clinic's actual care process.

What should staff do when check-in times out?

Preserve the intended patient and request context, verify whether the visit already exists and use the configured recovery route before creating another one.

Can a duplicate queue row simply be deleted?

First inspect attached notes, observations, orders and money records. Use an attributable correction process that preserves and reconciles affected work.

Sources and scope

Updated on . Original event-identity and recovery model with a fictional attempt count. The provider source illustrates a technical principle, not Bractis behaviour or a universal duplicate-prevention guarantee. No independent clinical or legal review is claimed.

  • Stripe: Idempotent requests

    Provider example of retry identity, with API-version differences; not a universal retention interval or Bractis implementation claim.

Try the workflow
with Bractis.

Ask Bractis to check in one synthetic appointment from two desks while a response is delayed, then test shared-phone relatives and a legitimate new episode. Reconcile stored visits and attached work.

Request a clinic demo