An immunisation register is a denominator, not a list
Most hospitals that say they run immunisation services are actually running a vaccination counter. A child arrives, a vaccine is given, a card is stamped and a bill is raised. That is a transaction log. A register is something different: it knows which children belong to your service, what each of them is due, and which of them have not come. The difference between the two is whether you can name the children who are overdue, and almost no transaction log can.
The distinction is worth labouring because it determines the entire system design. A transaction log needs a product, a price and a date. A register needs a defined cohort, a schedule model, a due-date engine and a follow-up workflow. Building the second on top of the first is possible but it means going back and answering the question you skipped, which is who counts as your population.
Define that cohort explicitly and write the rule down. A common and workable definition is every child born in the facility plus every child who has received at least one dose at the facility, remaining in the cohort until a defined age or until formally transferred out. Whatever you choose, the rule has to be applied by the system rather than by judgement, because a cohort that different staff interpret differently produces coverage figures that mean nothing.
Modelling the schedule so due dates compute themselves
The schedule is the heart of the system and it should be configuration, not code. Model each vaccine as a series of doses, each dose with a due age, a minimum interval from the previous dose, and a valid age window. Once that structure exists, the due date for any child at any time is a computation from date of birth and dose history rather than something a nurse works out at the counter with a card and a calendar.
You will need to hold more than one schedule. The National Immunisation Schedule under the Universal Immunisation Programme is the baseline and is what public reporting expects, while many private paediatric practices follow the schedule published by the Indian Academy of Pediatrics, which includes additional vaccines. Rather than forcing a choice, model both and let the child's record indicate which is being followed, with the ability to record a dose given elsewhere against either.
Build in the exceptions from the start because they are not rare. Preterm infants, children starting late and needing a catch-up sequence, children with a dose recorded from another provider, and children with a documented contraindication all need to be representable. A schedule engine that cannot express a catch-up sequence will be abandoned by the paediatricians within weeks. The clinical rules governing catch-up and contraindication belong to national guidance and to the treating clinician, and the system should record decisions rather than make them.
What each dose definition needs to carry
- Recommended age or interval from the previous dose in the series
- Minimum valid interval and the earliest and latest acceptable ages
- Which schedule it belongs to, so multiple schedules can coexist
- Route and site convention for recording, without prescribing clinical technique
- Whether a dose recorded from an outside provider satisfies this position in the series
Due, overdue and dropped-out lists that someone actually works
A recall engine produces three lists and each needs a different owner and a different rhythm. The due list is forward-looking and drives reminders. The overdue list is a short window past the due date and drives active follow-up calls. The dropped-out list covers children who have missed by a long margin and needs a decision rather than a call, because at some point continuing to chase a family who has moved cities is not a good use of anyone's time.
Assign the work to a named person with time allocated for it. In practice this is usually an immunisation counter nurse or a patient coordinator, working the overdue list for a fixed hour a day. Without that allocation the lists generate faithfully and nobody opens them, which is the most common way immunisation recall projects fail. The software is rarely the problem; the unassigned hour is.
Give the worker a way to record the outcome of every attempt. Contacted and appointment booked, contacted and declined, number not reachable, moved away, receiving vaccines elsewhere. Those outcomes are what let you distinguish a coverage problem from a data problem, and after three months they will tell you that a meaningful share of your overdue list is simply wrong phone numbers, which is a registration desk fix rather than a clinical one.
Contact outcomes to capture on every recall attempt
- Reached and appointment scheduled, with the date offered
- Reached and declined, with the stated reason recorded verbatim
- Number unreachable or incorrect, flagged back to registration for correction
- Family relocated or now vaccinating elsewhere, with a transfer-out marker
- No response after the agreed number of attempts, moving the child to the dropped-out list
Reminders, consent and the rules that govern them
Automated reminders make recall affordable, and they are the part of the system most likely to land you in a compliance problem. Health-related messages to a parent about a child are personal data processing under the DPDP Act 2023, which means you need a lawful basis, a purpose stated in language the parent understands, and a working way to withdraw. Consent captured once at registration in a tick box nobody read will not survive scrutiny.
The channel rules are separate and equally real. Commercial and transactional messaging over SMS in India runs through the DLT registration regime under the TRAI telecom commercial communications regulations, with registered headers and approved templates. WhatsApp Business messaging requires its own opt-in and template approval. Neither of these is optional, and both take longer to arrange than teams expect, so start them early in the project rather than the week before go-live.
Design the message content with as much care as the plumbing. Keep it short, in the language the family registered in, state the child's first name and the vaccine due rather than a clinical code, and include a way to reschedule that does not require a phone call during working hours. Avoid anything that reads as a marketing message. A reminder that feels like an advertisement gets the whole sender identity blocked, which takes out the reminders for every other family too.
“We had the recall lists running for eight months before we sorted out the messaging templates properly. The lists were accurate the whole time and coverage barely moved, because a nurse ringing two hundred families a month is not a channel.”
Batch, vial and cold chain on the same record
Every dose administered needs to carry the vaccine name, manufacturer, batch number and expiry, recorded at the point of administration rather than reconstructed later. This is what makes a recall of a specific batch a query rather than a crisis. If a manufacturer or the regulator issues an alert on a batch, a unit with batch-level records can produce the list of affected children in minutes; a unit without them has to review every file.
The cold chain record sits alongside and needs to be equally disciplined. Twice-daily temperature readings for each ice-lined refrigerator and deep freezer, with the reading, the time and the person recording it. Vaccine vial monitor status checked and recorded at issue. Alarm and excursion events logged with the action taken and the decision on the affected stock. Many units now run continuous electronic temperature logging, which removes the transcription burden but does not remove the requirement to review and act on excursions.
Connect the two records. A temperature excursion is only actionable if you can identify which vials were in that unit during the excursion window and which children received doses from them. That link, from storage unit to vial to administered dose to child, is the whole point of the exercise, and it is usually the piece that is missing when a hospital discovers it needs it.
Records that must exist for every administered dose
- Vaccine, manufacturer, batch number and expiry captured at administration
- Storage unit the vial was drawn from, linking the dose to the cold chain log
- Vaccine vial monitor status at the point of issue
- Administering staff member, date and time
- Any adverse event following immunisation reported and linked to the dose record
Measuring coverage, dropout and what the numbers hide
Two measures carry most of the meaning. Coverage is the proportion of your defined cohort that has received a given dose by a given age. Dropout is the proportion who received an early dose in a series but not a later one, and it is the more actionable of the two because it isolates children you have already reached and then lost. A unit with high coverage and high dropout has a follow-up problem; a unit with low coverage and low dropout has an access or registration problem.
Compute both against the register cohort and be explicit that this is a facility denominator, not a population one. A private hospital cannot claim area coverage, and it should not try. What it can say honestly is what proportion of the children it considers its own are up to date. Publishing that with the cohort definition attached is credible; publishing an unqualified coverage percentage is not.
Watch the number that is easiest to improve dishonestly. Removing hard-to-reach children from the cohort raises coverage instantly, and transfer-out markers are the mechanism by which that happens, usually without anyone intending it. Audit a sample of transfer-out records each quarter and confirm that a real contact established the reason. A register that quietly discards its difficult cases will report excellent figures and mean nothing at all.
What the register cannot do for you
It cannot fix a bad phone number, and a large share of what looks like vaccine hesitancy is a registration data problem wearing a clinical costume. Validating the parent contact number at the point of birth registration, and reconfirming it at each visit, does more for recall performance than any refinement of the reminder logic. Make the confirmation a one-tap action at the counter rather than a form field, or it will not happen.
It also cannot substitute for the public system record. Children who receive doses at a government session, an outreach camp or another provider need those doses recorded in your register, and the only reliable route is asking to see the immunisation card at every visit and entering what is on it. The national digital immunisation platform is progressively improving portability, but until any given facility is genuinely exchanging records, the card in the parent's bag remains the source of truth.
Finally, it will not tell you whether your service is any good. Coverage measures whether children came back. It says nothing about waiting times at the counter, whether the parent understood what was given, whether the adverse event advice was explained, or whether the card was filled in legibly. Those are experience and quality questions, and they need their own measurement, usually a short exit conversation with a handful of parents each week.
