Revisiting the Table 2 fallacy: A motivating example examining preeclampsia and preterm birth.

BACKGROUND: A "Table Fallacy," as coined by Westreich and Greenland, reports multiple adjusted effect estimates from a single model. This practice, which remains common in published literature, can be problematic when different types of effect estimates are presented together in a single table. The purpose of this paper is to quantitatively illustrate this potential for misinterpretation with an example estimating the effects of preeclampsia on preterm birth. METHODS: We analysed a retrospective population-based cohort of 2 963 888 singleton births in California between 2007 and 2012. We performed a modified Poisson regression to calculate the total effect of preeclampsia on the risk of PTB, adjusting for previous preterm birth. pregnancy alcohol abuse, maternal education, and maternal socio-demographic factors (Model 1). In subsequent models, we report the total effects of previous preterm birth, alcohol abuse, and education on the risk of PTB, comparing and contrasting the controlled direct effects, total effects, and confounded effect estimates, resulting from Model 1. RESULTS: The effect estimate for previous preterm birth (a controlled direct effect in Model 1) increased 10% when estimated as a total effect. The risk ratio for alcohol abuse, biased due to an uncontrolled confounder in Model 1, was reduced by 23% when adjusted for drug abuse. The risk ratio for maternal education, solely a predictor of the outcome, was essentially unchanged. CONCLUSIONS: Reporting multiple effect estimates from a single model may lead to misinterpretation and lack of reproducibility. This example highlights the need for careful consideration of the types of effects estimated in statistical models.

Dental hygiene students' participation as subjects in a randomized controlled trial: A full evidence-based dentistry experience.

PURPOSE: Demonstrate that dental hygiene students' participation in a randomized controlled trial comparing the plaque-removing capability of two toothbrushes provides a fuller understanding of the factors affecting the potential application of research to practice. METHODS: All students (N = 18) in a baccalaureate dental hygiene class were engaged in the design of a randomized controlled trial using the Consolidated Standards of Reporting Trials (CONSORT) standard and then participated as subjects in a staggered, repeated measures trial using plaque removal as the dependent variable and brush type, brush head wear, and time in study as independent variables. A debriefing of student participation and lessons learned from analyzing the results was conducted. RESULTS: The study found statistically significant differences in plaque removal capability. Brush type accounted for only 4% of the variance, while measures of brush head wear were inconsistent, and time in the study ("experimental fatigue") accounted for the most variance (9%). Students recognized and confirmed by their personal experiences that research that fails to focus on variance can create an overly optimistic impression of research effectiveness. There was strong agreement that subjects/patients vary widely and that performance depends on multiple factors. CONCLUSIONS: Dental hygiene students who participated as subjects in a randomized controlled trial comparing toothbrushes for plaque removal capacity felt that full analysis to account for all sources of variance and estimate the magnitude of measures of effect add to the value of reported research. Variation across patients is important in practice, if often overlooked as an "error" in the literature.

Ultraprocessed Food and Cardiovascular Risk: Estimating the Number Needed to Harm in an Unfamiliar Situation.

The Number Needed to Harm (NNH) statistic is a measure of effect size. It is defined as the number of patients who need to be treated for one additional patient to experience an adverse outcome. The NNH is conventionally calculated in the context of a randomized controlled trial. This article explains how the NNH can be estimated and understood for a lifestyle behavior in the context of an observational study in which the outcome was described using an uncommon unit. The lifestyle behavior, here, was the intake of ultraprocessed food and the outcome was stated as the number of events per 1000 person-years. The NNH can be estimated from the data provided, expressed in different ways, and converted into a form that is relevant to clinical practice.

Need for and Practical Interpretations of the Person-Year Construct in Neuropsychiatric Research.

In observational studies, groups of interest may be carved out of predictors of interest. Thus, for example, if cardiovascular (CVS) health at age 50 years is the predictor of interest for dementia as the long-term outcome, groups of interest could comprise persons with poor, intermediate, and optimal CVS health at age 50. These groups would almost certainly be unbalanced in terms of sample size and duration of follow-up when incident dementia is assessed. The present article is a companion to the previous article in this column; it explains why the duration of follow-up needs to be factored into the sample size in each group, making person-years rather than persons as the unit for risk assessment. Next, this article explains how to reverse calculate an estimate of the number needed to treat (NNT) statistic in such situations. Finally, this article explains why the NNT so estimated is an approximation and not a true estimate of the NNT in the population.