This function calculates the probabilistic version of the vaccination timeliness analysis: the distribution of vaccinations by number of days elapsed since the start of the timeliness window (age_schedule_low), for each dose and further disaggregating by geographic level, using data from the electronic immunization registry (EIR). Doses are never pooled together, since each dose has its own timeliness window. It returns, for each day, the empirical probability density (pdf), i.e. the percentage of records vaccinated on that day, and the empirical cumulative distribution (cdf), i.e. the percentage of records vaccinated up to that day.
Usage
vt_distribution(
data.EIR,
data.schedule,
geo_level = "ADM0",
vaccines = NULL,
birth_cohorts = NULL
)Arguments
- data.EIR
Data frame. A data frame containing individual vaccination records. See
immunABC.EIRfor expected structure.- data.schedule
Data frame. A data frame defining the vaccination schedule. See
immunABC.schedulefor expected structure.- geo_level
Character. The geographic level to aggregate results by. Must be "ADM0", "ADM1" or "ADM2". If not specified, the default is "ADM0".
- vaccines
Character (optional). A character vector specifying the doses to include in the analysis. If
NULL(default), all doses indata.scheduleare included.- birth_cohorts
Numeric (optional). A vector specifying the birth cohort(s) for which the distribution should be calculated. If
NULL(default), all available cohorts are pooled together. Note that the most recent cohorts may not have had time to receive late vaccinations, which biases the distribution towards faster vaccination; consider excluding cohorts with incomplete follow-up. Specifying cohorts also excludes records with nodate_birth, son_missingmay be undercounted.
Value
A data frame with one row per dose (and geographic area, if any) and day since the start of the timeliness window, with the following columns:
dose: the dose.day: days since the start of the timeliness window (age_schedule_low).age: age at vaccination in days (day + age_schedule_low).n: number of records vaccinated on that day.pdf: percentage of records vaccinated on that day.cum_n: cumulative number of records vaccinated up to that day.cdf: cumulative percentage of records vaccinated up to that day.n_at_risk: number of records not yet vaccinated at the start of that day (population "at risk").total: number of vaccinations in the EIR for thatdoseand geographic area, excludingn_earlyandn_missing.n_early: number of records vaccinated before the timeliness window, excluded fromtotal.n_missing: number of records with a missing date, excluded fromtotal.window_end: last timely day (age_schedule_high - age_schedule_low).age_schedule_low: age in days for the lower limit of the timeliness window for thatdose, comes fromdata.schedule.
Details
Day 0 corresponds to age_schedule_low, so records vaccinated before the timeliness window (n_early) are excluded from the distribution and from its denominator (total), as are records with a missing date (n_missing). Because of this, the cdf at window_end is not equal to the TIMELY rate from vt_rate, which keeps early records in its denominator. Also note that, since only vaccinated records are analyzed, the cdf always reaches 100%, which overestimates the probability of vaccination of the whole target population.