Epidemiological methods: about time

Abstract

Epidemiological studies often produce false positive results due to use of statistical approaches that either ignore or distort time. The three time-related issues of focus in this discussion are: (1) cross-sectional vs. cohort studies, (2) statistical significance vs. public health significance, and (3), how risk factors "work together" to impact public health significance. The issue of time should be central to all thinking in epidemiology research, affecting sampling, measurement, design, analysis and, perhaps most important, the interpretation of results that might influence clinical and public-health decision-making and subsequent clinical research.

Keywords: effect sizes; mediators; moderators; risk factors; statistical and clinical significance.

Figure 1.

Comparison of two hypothetical survival curves for the subpopulations with RF = 1 and RF = 0 (non-proportional hazards).

Figure 2.

ROC curve comparing survival in the “high” and “low” risk groups.

Figure 3.

Equipotency curves comparing the locus of all points (p₀, p₁) with p₁ > p₀, the ROC plane with Odds Ratio = p₁(1 − p₀)/[(1 − p₁)p₀] and NNT = 1/(p₁ − p₀) = 3. (The Figure 3 here selected to be equal to (OR^1/2 + 1)/(OR^1/2 − 1)).

Figure 4.

Comparison of Y = (OR^1/2 − 1)/(OR^1/2 + 1) with 1/NNT for various follow-up times for the survival curves shown in Figure 1.