Match Cohort to Reduce Observational Bias — match

This function builds pairs of vaccinated and unvaccinated individuals with similar characteristics. The function relies on the matching algorithm implemented in the package {MatchIt}, setting, by default, method = "nearest", ratio = 1, and distance = "mahalanobis". Exact and near characteristics are accepted, passed in the parameters exact and nearest, respectively. The parameter nearest must be provided together with the calipers as a named vector (e.g., nearest = c(characteristic1 = n1, characteristic2 = n2), where n1 and n2 are the calipers). The default matching method of the function is static. This means that pairs are matched once, without taking into account their vaccination, censoring, and outcome dates. After this, the pairs whose exposition time do not coincide are removed to avoid negative time-to-events. The function returns a matched and adjusted by exposition cohort, with information of the beginning of follow-up period of pairs (t0_follow_up), corresponding to the vaccination date of the vaccinated individual, the individual time-to-event (time_to_event) and the outcome status (outcome_status), both taking into account the right-censoring dates. Pairs are censored if the vaccinated or unvaccinated partner was previously censored (i.e., if censoring_date_col is informed) and the censor occurs before their outcomes. Rolling calendar matching method will be included in future releases.

Usage

match_cohort(
  data_set,
  outcome_date_col,
  censoring_date_col,
  start_cohort,
  end_cohort,
  method = "static",
  nearest = NULL,
  exact = NULL,
  immunization_date_col = "immunization_date",
  vacc_status_col = "vaccine_status",
  vaccinated_status = "v",
  unvaccinated_status = "u"
)

Arguments

data_set: data.frame with cohort information.
outcome_date_col: Name of the column that contains the outcome dates.
censoring_date_col: Name of the column that contains the censoring date. NULL by default.
start_cohort: Start date of the study.
end_cohort: End date of the study.
method: Method to match the cohort. Default is static.
nearest: Named vector with name(s) of column(s) for nearest matching and caliper(s) for each variable (e.g., nearest = c("characteristic1" = n1, "characteristic2" = n2), where n1 and n2 are the calipers). Default is NULL.
exact: Name(s) of column(s) for exact matching. Default is NULL.
immunization_date_col: Name of the column that contains the immunization date to set the beginning of the follow-up period (t0_follow_up). Default is immunization_date.
vacc_status_col: Name of the column containing the vaccination. Default is vaccine_status.
vaccinated_status: Status assigned to the vaccinated population. Default is v.
unvaccinated_status: Status assigned to the unvaccinated population. Default is u.

Value

object of the class match. List with results from static match: match: data.frame with adjusted cohort, summary: matching summary, balance_all: balance of the cohort before matching, balance_matched: balance of the cohort after matching.

Four columns are added to the structure provided in data_set: subclass: ID of, matched pair, t0_follow_up: beginning of follow-up period for pair, time_to_event: time to event, and outcome_status: outcome status (1:positive, 0: negative).