Z-scores and the birthweight paradox

摘要: Summary Investigators have long puzzled over the observation that low-birthweight babies of smokers tend to fare better than non-smokers. Similar observations been made with regard factors other smoking status, including socio-economic race and parity. Use standardised birthweights, or birthweight z-scores, has proposed as an approach resolve crossing curves is hallmark so-called paradox. In this paper, we utilise directed acyclic graphs, analytical proofs extensive simulation study consider use z-scores their effect on statistical analysis. We illustrate causal questions implied by inclusion in models, utility models include address those questions. Both analytically through a show neither nor z-score adjustment may be used for decomposition. The yields unbiased estimate total effect, even when collider-stratification would adversely impact estimates from birthweight-adjusted models; however, could estimated more directly unadjusted model. does not add additional information beyond models. Thus, ability successfully paradoxical mortality due alteration parameter being (total effect), rather confounding decomposition factors.