Examiner Hot-Seat Flashcards
(15 cards)
Why did the authors choose a prospective cohort instead of an RCT?
An RCT of 25 K+ diabetics for 10 y is ethically, logistically, and financially impossible; a well-phenotyped cohort still captures incident CVD while allowing real-world diet variation.
How was dietary mis-reporting minimised?
Five repeated 24-h Oxford WebQ recalls (2009-2012) were averaged; extreme energy intakes were excluded and diet validity of WebQ is biomarker-validated.
Why exclude events in the first 2 years during sensitivity analyses?
To cut reverse-causation—people might change diet due to pre-existing, undiagnosed CVD symptoms. Results were unchanged, supporting causality.
Overall PDI showed no effect—interpret that.
Quantity of plant food alone isn’t enough; the risk signal is driven by quality. Mixing healthy and ultra-processed plants in one score diluted associations.
Name two key dietary mediators of the uPDI–CVD link.
Low whole-grain intake (explained 36 % of harm in diabetes) and high SSB intake (15 % in pre-DM).
Which biomarker mediated the greatest proportion of uPDI harm?
Serum cystatin C—15 % of excess risk in pre-DM and 44 % in diabetes, implicating the kidney-cardio axis.
Jadad score is only 1/5—does that invalidate the study?
No. Jadad penalises non-randomised designs; it doesn’t assess observational quality. Newcastle–Ottawa (9/9) and Downs-&-Black (24/28) confirm high methodological rigour.
Biggest residual confounder after multivariable adjustment?
Unmeasured diet quality change over 10 y; only baseline recalls were used, so later improvements or regressions aren’t captured.
How might 70 % attrition in pre-DM bias findings?
Healthy-volunteer bias: those who completed WebQs are likelier health-conscious, which could underestimate true harm of uPDI in the broader population.
Portion size isn’t captured—why is that important?
Quintile scoring ranks frequency but not grams; two small SSBs could equal one mega-bottle—limits translation into concrete serving advice.
Why is hPDI significant in pre-DM but null in DM?
Earlier disease stage retains metabolic plasticity; in established diabetes renal and vascular damage (reflected by high cystatin C) may blunt diet benefits.
Generalise these results to Afro-Caribbean populations?
Caution: cohort is >90 % White British; prior US studies found no PDI benefit in non-Hispanic Blacks, so replication in diverse groups is needed.
Difference between decomposition and mediation analyses here?
Decomposition partitions diet-score effect into food groups (e.g., whole-grains), while mediation quantifies biomarker pathways (e.g., cystatin C) explaining the diet–CVD link.
What is the Townsend Deprivation Index and why adjust for it?
A UK postcode-level socioeconomic score; deprivation correlates with both diet quality and CVD, so adjusting removes socioeconomic confounding.
If you had £500 K to translate these findings, what intervention would you test?
A pragmatic pre-DM trial swapping SSBs/refined grains for subsidised whole-grains & legumes, tracking cystatin C and CVD risk markers over 2 years.
