Conditional Logistic Regression in jamovi

Matched case-control studies with the condLogReg module

In a matched case-control study each case is compared with one or more controls chosen to resemble it on factors such as age, sex or place of residence. The matched sets are part of the design, and the analysis has to keep them: comparisons are made within sets, never across them. Conditional logistic regression does exactly that. It estimates odds ratios from the contrast between each case and its own controls, and the set-specific baseline risks, which a matched design cannot estimate, drop out of the likelihood (Breslow et al. 1978; Breslow and Day 1980).

The condLogReg module brings this analysis to jamovi. It is a clone of jmv’s Binomial Logistic Regression, with the same model builder, reference levels, fit measures and coefficient tables, plus a Matching Variable that identifies the matched sets. The model is fitted by the conditional likelihood, as a Cox model stratified by matched set: the same fit as R’s survival::clogit (Gail et al. 1981; Therneau and Grambsch 2000).

In jamovi

Analyses ▸ Regression ▸ Logistic Regression ▸ Conditional (Matched)

Box What goes in it
Dependent Variable case/control status, a factor with two levels; its reference level is the control
Matching Variable the identifier of the matched set (stratum, pair, set number)
Covariates numeric exposures and confounders
Factors categorical exposures and confounders

The rest of the panel is the one jmv users know: Model Builder (blocks of predictors, interactions), Reference Levels, Assumption Checks (collinearity), Model Fit (deviance, AIC, BIC, pseudo R², overall test) and Model Coefficients (omnibus likelihood ratio tests). Several parts are new or laid out differently:

  • Matching: the conditional likelihood, exact or approximate;
  • Descriptives: the matched-set table, and a descriptive table of every variable by case/control status, with its missing values;
  • Model Coefficients: one row per level, with the reference level at OR = 1 and the odds ratio, its confidence interval and p for the others; univariable beside multivariable odds ratios, a trend test and a forest plot on request;
  • Options: Show notes, the explanations under each table (off by default).

A first result, with the module’s bundled data set (Infertility after abortion, R’s infert): 248 women in 83 matched sets, each infertile case matched with up to two fertile controls on age, parity and education (Trichopoulos et al. 1976).

Matched Sets
Per set = Cases Controls Sets Observations
1 1 1 2
1 2 82 246
Model Coefficients - case
OR Lower 95% CI Upper 95% CI p
spontaneous per unit 7.285 3.651 14.536 < .001
induced per unit 4.092 2.018 8.298 < .001

Each previous spontaneous abortion multiplies the odds of secondary infertility by 7.29, and each induced abortion by 4.09, comparing women of the same age, parity and education.

What is in this tutorial

  • Method: why matched data need a conditional analysis, the conditional likelihood, its equivalence with a stratified Cox model, and what can and cannot be estimated.
  • Reading the output: every table and option of the analysis, how each number is computed, and what changed from jmv’s logistic regression.
  • Use cases in jamovi: three studies analysed step by step, with their data to download. Two of them ship with the module in jamovi’s Data Library, with a third, simulated one.

Install

The module is sideloaded from the .jmo files on the repository’s Releases page, one per jamovi series and CPU: in jamovi, Modules ▸ jamovi library ▸ Sideload. It needs nothing beyond the packages bundled with jamovi (jmvcore, R6, ggplot2 and survival).

Back to top

References

Breslow, N. E., and N. E. Day. 1980. Statistical Methods in Cancer Research. Volume I: The Analysis of Case-Control Studies. IARC Scientific Publications 32. International Agency for Research on Cancer.
Breslow, N. E., N. E. Day, K. T. Halvorsen, R. L. Prentice, and C. Sabai. 1978. “Estimation of Multiple Relative Risk Functions in Matched Case-Control Studies.” American Journal of Epidemiology 108 (4): 299–307. https://doi.org/10.1093/oxfordjournals.aje.a112623.
Gail, Mitchell H., Jay H. Lubin, and Lawrence V. Rubinstein. 1981. “Likelihood Calculations for Matched Case-Control Studies and Survival Studies with Tied Death Times.” Biometrika 68 (3): 703–7. https://doi.org/10.1093/biomet/68.3.703.
Therneau, Terry M., and Patricia M. Grambsch. 2000. Modeling Survival Data: Extending the Cox Model. Springer. https://doi.org/10.1007/978-1-4757-3294-8.
Trichopoulos, D., N. Handanos, J. Danezis, Anna Kalandidi, and Victoria Kalapothaki. 1976. “Induced Abortion and Secondary Infertility.” British Journal of Obstetrics and Gynaecology 83 (8): 645–50. https://doi.org/10.1111/j.1471-0528.1976.tb00904.x.