Matched case-control design

Description

Declares an individually or frequency matched case-control sample: each case is matched to one or more controls sharing the values of the matching variable(s). The design-faithful analysis is conditional logistic regression (conditional maximum likelihood), which conditions on the matched-set totals and so removes the matching-variable nuisance parameters; fitting matched-set indicators by ordinary logistic regression biases the odds ratio and is never used.

Usage

matched_cc(strata, ratio = NULL, prevalence = NULL, prevalence_n = NULL)

Arguments

strata A non-empty character vector naming the column(s) that identify the matched sets (e.g. “set”, or c(“age_grp”, “sex”) for frequency matching).
ratio NULL or a single whole number >= 1. The number of controls matched per case (m:1). Optional metadata; the conditional likelihood does not require a fixed ratio.
prevalence NULL or a single number in (0, 1): the source-population prevalence q0, required for the case-control-weighted (“ccw_*“) marginal estimators. Optional for the (default) conditional logistic OR. For a case-control-weighted marginal effect on matched data the matching variable(s) must be included in confounders (the matched controls are not a representative population control sample, so the marginal estimator adjusts for the matching variable rather than conditioning on the matched sets; Rose & van der Laan 2009 — matching may cost efficiency relative to an unmatched CCW analysis).
prevalence_n NULL or a whole number giving the cohort size q0 was estimated from (makes q0 estimated, adding a variance term).

Value

A matchatr_design object of type “matched_cc” carrying the strata, ratio, and (optionally) the prevalence.

References

McNemar Q (1947). Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika 12(2):153-157.

Breslow NE, Day NE (1980). Statistical Methods in Cancer Research, Volume 1: The Analysis of Case-Control Studies. IARC Scientific Publications No. 32.

Rose S, van der Laan MJ (2009). Why match? Investigating matched case-control study designs with causal effect estimation. The International Journal of Biostatistics 5(1):Article 1.

See Also

unmatched_cc(), nested_cc(), matcha()

Other design constructors: case_cohort(), counter_matched(), nested_cc(), two_phase(), unmatched_cc()

Examples

library("matchatr")

matched_cc(strata = "set")
<matchatr_design>
 Type:       Matched case-control
 Strata:     set
 Weights:    none
matched_cc(strata = c("age_grp", "sex"), ratio = 2)
<matchatr_design>
 Type:       Matched case-control
 Strata:     age_grp, sex
 Ratio:      2:1
 Weights:    none
matched_cc(strata = "set", prevalence = 0.02) # for a marginal CCW contrast
<matchatr_design>
 Type:       Matched case-control
 Strata:     set
 Prevalence: 0.02 (q0)
 Weights:    case-control (q0)