Skepticism
The headline: “Drinking coffee prevents heart disease” ignores these important aspects of a study.
Ignores confounding and study design limitations like measurement error and dose-response.
This criterion requires exposure to occur before outcome.
Temporality
This study design follows people forward in time to track new cases of disease.
Prospective cohort
This is the DHHS aerobic PA guideline for adults.
150-300 minutes of moderate-intensity (or 75-150 minutes of vigorous) activity.
This is the definition of prevalence.
The proportion of a population with an outcome at a specific time.
Rewrite the headline "Higher step counts are associated with lower risk of early death" to be more accurate and cautious (demonstrate that you know what COULD be included not what reflects reality).
Including details such as study design, study population, or measurement type.
E.g., "In a prospective cohort of U.S. older adults, accelerometer-measured step counts were linked to lower mortality risk in a dose-response fashion."
In the acetaminophen–autism research, this criterion is violated if sibling studies find null results.
Consistency
This study design is strongest for establishing causality but is rarely feasible in epidemiology.
Randomized controlled trial (RCT)
Convert 30 min of brisk walking (4 METs) into MET-minutes.
120 MET-min
This is the definition of incidence.
The number of new cases in a population over a defined period of time.
These biases might distort a study’s results when reported in the news.
Selection bias, recall bias, measurement error, confounding.
These are typical hazard ratios for the PA–T2D and smoking–T2D relationships, and contribute to this aspect of Hill’s criteria?
PA–T2D: ~0.7 (protective)
Smoking–T2D: ~1.1–1.4 (harmful)
Contributes to: Strength of association
This study design compares people with and without a disease at one point in time.
Case-control study
This PA measure is prone to recall bias.
Self-report
This is the interpretation of a RR of 0.5 for a study examining the association between meeting PAGs and the risk of incident T2D.
People meeting PAGs have half the risk of developing diabetes compared with those who don’t
This is the difference between relative and absolute risk. (Explain why it matters).
Relative risk compares proportional change (e.g., 50% reduction), while absolute risk gives the actual numbers (e.g., 2% down to 1%). Absolute risk shows real-world impact.
This Hill’s criterion is most affected by publication bias.
Consistency
Sibling cohort studies have this advantage compared with standard cohort studies.
They control for shared genetics and environment (family-level confounders).
Compare and contrast accelerometers and surveys for measuring PA in an after-school PA intervention.
Accelerometers provide objective data but are heavily influenced by wear-time and don't capture type of activity well. Surveys provide context for type but are prone to recall and social desirability bias.
This is an explanation of the odds ratio from a study of early-onset colon cancer: researchers enrolled 120 cases and 360 controls. Among cases, 40 were active and among controls, 240 were active.
If active is considered the reference: (80/40)/(120/240) = 4.0
If inactive is considered the reference:(40/80)/(240/120) = 0.25
This headline is more likely to be causal overclaim: “Exercise reduces T2D risk” or “Exercise prevents T2D.” Explain why.
"Exercise prevents T2D.” It implies that exercise is sufficient for prevention, while “reduces risk” is more accurate since T2D can occur in individuals who are physically active.
This aspect of Hill's criteria is most weakened by the fact that acetaminophen use has decreased while autism prevalence has risen. (Must be able to explain why).
Because the exposure and outcome trends move in opposite directions, making a causal link less coherent.
This is a limitation of biomarker studies in prenatal acetaminophen research. This limitation is also demonstrated by increased all-cause mortality risk in individuals taking beta-blockers.
Biomarkers may reflect underlying illness or indication for use (confounding by indication).
Calculate total calories expended: 70-kg person running at 6 METs for 30 minutes.
~210 kcal (6 METs × 3.5 × 70 ÷ 1000 × 5 × 30)
This is the relative risk from a 10-year cohort, 40 of 600 active adults and 150 of 400 inactive adults develop CVD.(Must give an interpretation).
Inactive as reference: 40/(600x10) divided by 150/(400x10) = 0.178
Active as reference: 150/(400x10) divided by 40/(600x10) = 5.53