Presentation is loading. Please wait.

Presentation is loading. Please wait.

Harmonisation between cohorts: Holy Grail or Fool’s errand.

Similar presentations


Presentation on theme: "Harmonisation between cohorts: Holy Grail or Fool’s errand."— Presentation transcript:

1 Harmonisation between cohorts: Holy Grail or Fool’s errand

2

3 Some of the problems: Lots of genes with modest effects Ergo, lots of genes modifying environmental effects Ill-defined outcomes Reliance on proximal phenotypes (Usually) imprecise exposure measurement Size versus detail; Cost!

4 Bigger studies (combining cohorts) versus Better measurement

5 Combination likely to inflate misclassification Genetic Outcome Exposure …..therefore, do the gains in sample size compensate for these effects?

6 Increased power for improved allele classification is not great for relatively common alleles

7 OUTCOMES - ASTHMA Physician diagnosed – direct observation Physician diagnosed – parent reported Parent-reported symptoms –Many studies have ISAAC or equivalent –GA 2 LEN survey suggests many studies use idiosyncratic definitions of eczema & hay fever

8

9 What about sub-phenotypes?

10

11 Comparison of sRaw in children with no history of wheeze, and unconfirmed and confirmed wheeze.

12

13 OUTCOMES – INTERMEDIATE Pulmonary function measurements –Equipment, protocols, standards Assessment of allergy –SPT v Specific IgE Bronchial hyperresponsiveness –Method, data censoring, repeatability

14 Maternal anxiety score 32 weeks GA OR adj AsthmaOR adj Asthma + BHR 1 st quartile1 (reference) 2 nd quartile1.36 (1.09-1.71)1.35 (0.79-2.30) 3 rd quartile1.42 (1.14-1.77)1.55 (0.93-2.58) 4 th quartile1.65 (1.30-2.08)2.09 (1.24-3.53) Stratification of asthma by objective measures, e.g.

15 EXPOSURES Mostly proxy measures (questionnaires, derived variables, modelled) Where directly assessed, often focus of one study (diet, pollution, allergen, endotoxin) Repeat measures likely to reduce misclassification Can we compare different methods?

16

17

18 Outcome and exposure misclassification have major effect on power to detect interactions Some outcomes sufficiently similar between cohorts to allow combined analyses Even with ‘objective’ outcomes there are potential problems Repeat assessment of exposures can reduce effect of misclassification

19 The fool doth think he is wise, but the wise man knows himself to be a fool.


Download ppt "Harmonisation between cohorts: Holy Grail or Fool’s errand."

Similar presentations


Ads by Google