Why repeated meta-analyses can show very different results?
•Download as PPTX, PDF•
3 likes•4,353 views
In this slide deck I describe 2 typical reasons which may cause diverging results in repeated meta-analyses on the same subject
Recommended
More Related Content
What's hot
What's hot (20)
Viewers also liked
Viewers also liked (16)
Similar to Why repeated meta-analyses can show very different results?
Similar to Why repeated meta-analyses can show very different results? (20)
More from Reijo Laatikainen
More from Reijo Laatikainen (9)
Recently uploaded
Recently uploaded (20)
Why repeated meta-analyses can show very different results?
- 1. Why repeated meta-analyses and systematic reviews may produce different outcomes? Recent meta-analyses on fat as an example Registered dietitian, M.Sc, MBA Reijo Laatikainen www.twitter.com/pronutritionist Page 1
- 2. Why this slide deck? • Early 2014 a meta-analysis by Chowdhury et al. in Annals of Intern Medicine concluded that omega-6 fat intake is not associated with coronary heart disease in prospective cohorts • In contrast in August 2014 a meta-analysis by Farvid et al. in Circulation concluded that linoleic acid is associated with reduced risk of coronary heart disease both independently and in substitution model against saturated fat • In this slide deck I compare subtle differences in Chowdhury’s and Farvid’s meta-analyses and their effect on the overall findings • I only compare the representative analyses, ie. the reported fat intake and risk of coronary heart disease (CHD). Chowdhury’s other analyses on RCTs and circulating fatty acids are beyond the scope of this slide deck. 2
- 3. Meta-analyses Chowdhury et al. (2014) vs Farvid (2014) regarding linoleic acid (omega-6 FA) intake in prospective cohorts 3
- 4. High versus low intake of omega-6 fatty acid/linoleic and risk of coronary heart disease Chowdhury Risk of CHD (omega-6 FA intake as variable) 95 % Confidence interval 0.98 0.90-1.06 Farvid Chowdhury R et al. Association of dietary, circulating, and supplement fatty acids with coronary risk: a systematic review and meta-analysis. Ann Intern Med. 2014 Mar 18;160(6):398-406. Farvid M. et al. Dietary Linoleic Acid and Risk of Coronary Heart Disease: A Systematic Review and Meta- Analysis of Prospective Cohort Studies. Circulation. 2014 Aug 26. pii: CIRCULATIONAHA.114.010236. 4 Risk of CHD (Linoleic acid intake as variable) 95 % Confidence interval 0.85 0.78-0.92
- 5. 1. Re-use of the original studies differ between meta-analyses 5
- 6. Farvid used original categorization of fat intake but Chowdhury created new thirds Chowdhury – new ”thirds” Farvid –original categories Chowdhury R et al. Association of dietary, circulating, and supplement fatty acids with coronary risk: a systematic review and meta-analysis. Ann Intern Med. 2014 Mar 18;160(6):398-406. Farvid M. Dietary Linoleic Acid and Risk of Coronary Heart Disease: A Systematic Review and Meta- Analysis of Prospective Cohort Studies. Circulation. 2014 Aug 26. pii: CIRCULATIONAHA.114.010236. 6
- 7. You may lose the actual effect of very high or low intake if you transform quintiles/quartiles to thirds (Chowdhury) On the other hand, you don’t have same metrics between individual studies if you compare quintiles to thirds (Farvid) 7
- 8. 2. Meta-analyses include and exclude studies differently depending on the search strategy defined before literature search 8
- 9. Studies included re: linolic acid intake and risk of coronary heart disease 9 Studies Only in Chowdhury Studies Only in Farvid Included in both Chowdhury and Farvid MrFit* Glostrup KIHD Monica (F) Monica (M) Aric (F) Aric (M) FMC (F) FMC (M) WHS VIP Morgen NHS HPFS ATBC MDC (ie. Malmö study) *) MrFit was included also in Farvid’s CHD mortality analysis but not in overall analysis of CHD events
- 10. No wonder that results are different. Selection of studies is strikingly different. 10
- 11. 3. Only Farvid’s meta-analysis included fat substitution modelling 11
- 12. A 5% of energy increment in LA intake replacing energy from saturated fat intake was associated with a 9% lower risk of CHD events in Farvid’s meta-analysis Chowdhury Farvid Chowdhury R et al. Association of dietary, circulating, and supplement fatty acids with coronary risk: a systematic review and meta-analysis. Ann Intern Med. 2014 Mar 18;160(6):398-406. Farvid M. et al. Dietary Linoleic Acid and Risk of Coronary Heart Disease: A Systematic Review and Meta- Analysis of Prospective Cohort Studies. Circulation. 2014 Aug 26. pii: CIRCULATIONAHA.114.010236. 12
- 13. Conclusion • Chowdhury’s and Farvid’s meta-analysis have less studies in common than they do not have in common (5 vs 11 studies) • Using the original categorization of data vs calculating new perhaps more comparable categories may either highlight or dim the actual effects • It’s important to consider direct fat-by-fat substitutions. Increased intake of carbohydrate or certain protein qualities can mask real effects of saturated fat if fat is replaced by them rather than (certain) fatty acids. Fat subsititution was done only in Farvid’s meta-analysis • Researchers can intentionally or unintentionally manipulate the outcomes of meta-analyses by (subtly) changing research methods and words describing what is being done. 13
- 14. Wellcome aboard! http://twitter.com/pronutritionist http://www.facebook.com/pronutritionist http://www.pronutritionist.net Reijo Laatikainen, RD, MBA Images bought and licensed from BigStockPhoto. Snapshots from papers and sites referred to. Page 14