Objective
Studies of intersectionality are increasing to examine health inequalities. Different proposals for examining intersections have recently been published. One approach (1) considers models specified with 1st and all 2nd -order effects and another (2) the stratification based on multiple covariates; both categorize continuous covariates. A simulation study was conducted in order to review both methods with regard to correct identification of intersections, rate of false positive results, and generalizability to independent data compared to an established approach (3) of backward variable elimination according to Bayesian information criterium (BE-BIC).
Study design and setting:
Two basically different settings were simulated with 1000 replications: (1) comprised the covariates age, sex, body mass index, education, and diabetes in which no association was present between covariates and a continuous response and (2), comprising the same covariates, and a non-linear interaction term of age and sex, i.e., a non-linear increase in females above middle age formed the intersection of interest. The sample size (N = 200 to N = 3000) and signal to noise ratios (SNR, 0.5 to 4) were varied. In each simulated dataset bootstrap with replacement was used to fit the model to internal learning data and to predict outcomes using the fitted models in these data as well as the internal validation data. In both, the mean squared error (MSE) was calculated.
Results
In simulation setting 1, approaches 1/2 generated spurious effects in more than 90% of simulations across all sample sizes. In smaller sample size, approach 3 (BE-BIC) selected 36.5% the correct model, in larger sample size in 89.8% and always had a lower number of spurious effects. MSE in independent data was generally higher for approaches 1/2 when compared to 3. In simulation setting 2, approach 1 selected most frequently the correct interaction but frequently showed spurious effects (> 75%). Across all sample sizes and SNR, approach 3 generated least often spurious results and had lowest MSE in independent data.
Conclusion
Categorization of continuous covariates is detrimental to studies on intersectionality. Due to high model complexity such approaches are prone to spurious effects and often lack interpretability. Approach 3 (BE-BIC) is considerably more robust against spurious findings, showed better generalizability to independent data, and can be used with most statistical software. For intersectionality research we consider it more important to describe relevant intersections rather than all possible intersections.