On time, late, or caught up? Why static coverage hides vaccination trajectories

#Question

Why is final vaccination status not enough to describe programme performance?

#Answer in brief

Because vaccination is a time-dependent intervention. WHO emphasizes timely vaccination because protection delayed can leave a person susceptible during the period when protection was intended. Catch-up is a necessary part of a functioning immunization programme, but being eventually caught up is not epidemiologically identical to being protected on schedule.^[1]

A final coverage percentage can therefore hide very different trajectories.

#Same endpoint, different histories

Consider two synthetic birth cohorts of 1,000 children.

At age 24 months, 950 children in each cohort have received a target dose.

Coverage is 95% in both.

But:

• in Cohort A, 900 received it within the recommended window and 50 received it late;
• in Cohort B, only 650 received it on time and 300 received it after an intensive catch-up effort at 20–23 months.

The endpoint is identical. The period of susceptibility is not.

If disease risk is concentrated early, the cohorts have experienced materially different protection.

#Static coverage compresses time

A cross-sectional status typically classifies a person as:

• vaccinated;
• not vaccinated.

Longitudinal monitoring needs more states:

• eligible but not yet due;
• due;
• vaccinated on time;
• overdue;
• partially vaccinated;
• caught up;
• ineligible for a subsequent dose until interval requirements are met;
• status uncertain because records are incomplete.

The exact categories depend on the schedule. The methodological point is general: status is a trajectory, not only an endpoint.

#Catch-up is routine programme function

WHO defines catch-up vaccination as vaccinating people who have missed doses for which they are eligible and states that catch-up should be a continuous component of routine immunization, not only an emergency campaign.^[1][2]

That distinction changes monitoring.

A programme that performs catch-up well needs to know:

• who is overdue;
• for which dose;
• by how long;
• whether prior intervals are valid;
• whether the person can be reached;
• whether later records show successful recovery.

Counting only doses administered during a catch-up campaign cannot answer those questions.

#Dose sequences create state dependence

Multi-dose schedules are not a set of independent checkboxes.

Eligibility for dose 2 can depend on:

• receipt of dose 1;
• minimum interval;
• age;
• maximum or recommended timing;
• product-specific rules;
• prior valid records.

A person can therefore be:

• unvaccinated;
• started but incomplete;
• complete on time;
• complete late;
• overdue for the next valid dose;
• documented with an invalid interval requiring interpretation under programme rules.

The sequence itself carries information.

#Cohorts matter

Calendar-year monitoring can mix people at different stages.

A birth-cohort approach can better answer questions such as:

• What proportion received dose 1 by the recommended age?
• What proportion completed the primary series by age X?
• Among those late for dose 1, how many were eventually caught up?
• How long did overdue periods last?
• Which groups had recurrent delay?

The appropriate cohort definition depends on the programme, but it should be fixed before analysis.

#Missed opportunities are not identical to non-vaccination

WHO defines a missed opportunity for vaccination as a health-service contact by an eligible person that does not result in administration of one or more doses for which that person is eligible.^[3]

This concept adds a service-process dimension.

A person can be overdue without having any relevant health-care contact. Another can be overdue despite repeated contacts where vaccination could have occurred.

The intervention differs:

• access problem;
• recall problem;
• eligibility recognition problem;
• stock problem;
• workflow problem;
• hesitancy or refusal;
• documentation problem.

Treating all overdue people as one category loses the mechanism.

#Monitoring catch-up requires different fields

WHO's 2025 implementation guidance on monitoring catch-up recommends adapting routine tools so that vaccination activity can capture information relevant to age and catch-up monitoring, and proposes indicators for programme monitoring.^[4]

A useful longitudinal dataset may need:

• person identifier;
• birth date;
• residence;
• dose/product;
• administration date;
• source/provider;
• prior-dose history;
• current eligibility;
• catch-up classification;
• date the record entered the system.

The last field matters because a dose can have been administered on time but recorded late. Administrative delay should not be confused with vaccination delay.

#A synthetic dashboard problem

A regional dashboard reports that coverage increased from 87% to 94% in six months.

This sounds unambiguously positive.

But the underlying trajectories show:

• most improvement came from people who were 12–18 months overdue;
• new infants entering eligibility are still being vaccinated late at the same rate as before;
• a temporary campaign closed an old gap but routine timeliness did not improve.

The programme achieved real benefit: previously unprotected people received vaccination.

Yet the system has not necessarily fixed the process that creates new delay.

A mature dashboard should show both:

• stock of people caught up, and
• flow of newly overdue people.

#Useful trajectory measures

Depending on the schedule, measures can include:

#On-time coverage

Proportion vaccinated within a defined recommended window.

#Median or distribution of delay

Among late vaccinations, how long after due date did administration occur?

#Catch-up completion

Among people overdue at a defined baseline, what proportion later received the missing dose?

#Dropout between doses

Among those who start a sequence, what proportion fail to receive later required doses within the expected interval?

#Zero-dose or no-record state

Who has no valid evidence of starting the schedule?

#Recurrence

Do the same populations repeatedly fall behind across multiple doses?

These measures answer different operational questions.

#Time also changes denominators

The denominator for "currently overdue" is not the same as the denominator for "ever eligible during the year."

People enter and leave eligibility. They may migrate, age out of a reporting definition or receive doses elsewhere.

Longitudinal monitoring therefore needs clear time anchors and versioned eligibility logic.

The companion article Vaccination coverage is a fraction with a moving denominator develops the denominator problem in more detail.

#Catch-up success can coexist with routine weakness

This apparent paradox matters.

A programme can run an excellent catch-up campaign and still have poor routine timeliness.

Conversely, a system can have strong on-time routine delivery and still need catch-up for people who move into the jurisdiction or whose records are incomplete.

Catch-up should therefore be interpreted as a complementary programme function, consistent with WHO guidance, not as evidence by itself that routine delivery is strong or weak.^[2]

#Practical monitoring architecture

A useful design separates three layers:

1. schedule logic — what should happen and when;
2. person trajectory — what actually happened;
3. programme state — aggregate patterns requiring action.

The schedule logic must be versioned. If recommendations change, a historical trajectory should not silently be reclassified without preserving the rule that applied at the time.

#Survival-style thinking can clarify timeliness

Vaccination timing can be analysed as time to event. Survival-analysis approaches have been used specifically to quantify age-appropriate vaccination delay rather than collapsing status into a final coverage proportion.^[5]

For a cohort, the event is receipt of a valid dose. People who have not yet received it remain at risk of being unvaccinated at each age or time point.

This view avoids forcing every person into a single final status too early.

A curve of cumulative vaccination over age can reveal whether two populations with the same final coverage reached protection at different speeds; that is the practical advantage of retaining the timing distribution.^[5]

The method chosen must reflect the schedule and censoring process, but the conceptual benefit is simple: time is retained rather than collapsed.

#Backlog and flow should be monitored together

A catch-up programme often inherits a stock of people already overdue.

At the same time, new people continue to become due.

If the programme vaccinates 1,000 overdue people but 1,200 newly eligible people become overdue, the backlog can grow despite substantial activity.

Monitoring only doses delivered can therefore hide deterioration.

Useful measures include:

• overdue stock at the start of the period;
• newly overdue during the period;
• caught up during the period;
• remaining overdue;
• median delay among those caught up.

This separates recovery performance from routine prevention of new gaps.

#Record uncertainty deserves its own state

Some people appear unvaccinated because the information system is incomplete.

Classifying them immediately as true zero-dose or overdue cases can waste outreach and distort coverage.

A trajectory system should allow a temporary status uncertain state when prior records require reconciliation.

The operational objective then becomes resolving uncertainty quickly enough that it does not delay genuinely needed vaccination.

#Limitations and uncertainty

This article is schedule-agnostic. Specific definitions of timely, overdue, valid interval, series completion and catch-up eligibility must come from current national or authoritative vaccine-specific guidance. WHO resources cited here provide general programme principles rather than a substitute for jurisdiction-specific schedules. The synthetic examples are illustrative and do not report data from a real vaccination programme.^[4]