Unmatched case-control design

Description

Declares an independent (unmatched) case-control sample: cases and controls are drawn separately from the source population with no individual or frequency matching. This is the design behind the classical logistic / Mantel-Haenszel odds ratio and, when the marginal outcome prevalence q0 is supplied, the case-control-weighted marginal causal contrasts of Rose & van der Laan (2009).

Usage

unmatched_cc(prevalence = NULL, prevalence_n = NULL, strata = NULL)

Arguments

prevalence	NULL or a single number in (0, 1). The marginal outcome prevalence q0 in the source population. Required for case-control-weighted (“ccw_*“) estimators, which reweight the sample back to the population; optional for the conditional / classical odds-ratio estimators, which do not need it.
prevalence_n	NULL (default; q0 is treated as known) or a whole number giving the cohort size q0 was estimated from. When supplied, the case-control-weighted variance gains a term for q̂0’s sampling uncertainty (Var(q̂0) = q0 (1 − q0) / prevalence_n), and a fit records prevalence_known = FALSE.
strata	NULL or a non-empty character vector naming the column(s) to stratify on for the Mantel-Haenszel estimator (estimator = “mh”), e.g. “agegrp” or c(“agegrp”, “sex”). Several columns are crossed into a single stratifying factor. Used only by “mh”; the “logistic” estimator adjusts for covariates via confounders instead, and “mh” with no strata reduces to the crude single-table odds ratio.

Details

Weights are never stored as a data column. The weight_spec records the intended scheme (case-control weighting when q0 is present, otherwise none); the weighting layer realises it as observation weights into the estimation engine.

Value

A matchatr_design object of type “unmatched_cc” carrying the prevalence, the Mantel-Haenszel strata, and a weight_spec describing the case-control weighting scheme.

References

Prentice RL, Pyke R (1979). Logistic disease incidence models and case-control studies. Biometrika 66(3):403-411.

Robins J, Breslow N, Greenland S (1986). Estimators of the Mantel-Haenszel variance consistent in both sparse data and large-strata limiting models. Biometrics 42(2):311-323.

Rose S, van der Laan MJ (2008). Simple optimal weighting of cases and controls in case-control studies. The International Journal of Biostatistics 4(1):Article 19.

See Also

matched_cc(), nested_cc(), case_cohort(), matcha()

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

Examples

library("matchatr")

unmatched_cc()

<matchatr_design>
 Type:       Unmatched case-control
 Weights:    none

unmatched_cc(prevalence = 0.02)

<matchatr_design>
 Type:       Unmatched case-control
 Prevalence: 0.02 (q0)
 Weights:    case-control (q0)

unmatched_cc(prevalence = 0.02, prevalence_n = 50000) # q0 estimated from a cohort

<matchatr_design>
 Type:       Unmatched case-control
 Prevalence: 0.02 (q0)
 Weights:    case-control (q0)

unmatched_cc(strata = "agegrp")

<matchatr_design>
 Type:       Unmatched case-control
 Strata:     agegrp
 Weights:    none