Reporting genetic association studies: the strega statement

cacy in today’s study 2 Kahn SE. Glucose control in type 2 diabetes. Still worthwhile and worth is encouraging, but it is impossible to draw defi nitive pursuing. JAMA 2009; 301: 1590–92.
Ray KK, Seshasai SR, Wijesuriya S, et al. Eff ect of intensive control of clinical conclusions from such a phase II study. Experience glucose on cardiovascular outcomes and death in patients with diabetes mellitus: a meta-analysis of randomised controlled trials. Lancet 2009; with muraglitazar has shown that a good lipid and 373: 1765–72.
cient to predict clinical 4 Gaede P, Lund-Andersen H, Parving HH, Pedersen O. Eff ect of a multifactorial intervention on mortality in type 2 diabetes. N Engl J Med outcomes. It is a strength that cardiovascular events, 2008; 358: 580–91.
including heart failure, were adjudicated in SYNCHRONY 5 Balakumar P, Rose M, Ganti S, Krishan P, Singh M. PPAR dual agonists: are they opening Pandora’s box? Pharmacol Res 2007; 56: 90–98.
(two cases of heart failure, no myocardial infarctions), 6 Staels B, Fruchart JC. Therapeutic roles of peroxisome proliferator-activated but the sample size and short duration of the study do receptor agonists. Diabetes 2005; 54: 2460–70.
The FIELD study investigators. Eff ects of long-term fenofi brate therapy on cardiovascular events in 9795 people with type 2 diabetes mellitus Diabetes is a complex disease and agents of the PPAR (the FIELD study): randomised controlled trial. Lancet 2005;
366: 1849–61.
class could be promising owing to the multiple eff ects 8 Keech AC, Mitchell P, Summanen PA, et al. Eff ect of fenofi brate on the need for laser treatment for diabetic retinopathy (FIELD study): a randomised of these drugs and the development of selective PPAR controlled trial. Lancet 2007; 370: 1687–97.
modulators,15 but long-term safety is a major concern. 9 Rajamani K, Colman PG, Li LP, et al, on behalf of the FIELD study investigators. Eff ect of a fenofi brate on amputation events in people with type 2 diabetes mellitus (FIELD study): a prespecifi ed analysis of a the coming years will tell whether hopes raised by the randomised controlled trial. Lancet 2009; 373: 1780–88.
10 Dormandy JA, Charbonnel B, Eckland DJ, et al. Secondary prevention of SYNCHRONY study for aleglitazar are confi rmed by macrovascular events in patients with type 2 diabetes in the PROactive Study (PROspective pioglitAzone Clinical Trial In macroVascular Events): appropriate long-term clinical trials.
a randomised controlled trial. Lancet 2005; 366: 1279–89.
11 Fievet C, Fruchard JC, Staels B. PPARα and PPARγ dual agonists for the treatment of type 2 diabetes and the metabolic syndrome. Curr Opin Pharmacol 2006; 6: 606–14.
Diabetes and Endocrinology Department, Institut du Thorax, 12 Rubenstrunk A, Hanf R, Hum D, Fruchart JC, Staels B. Safety issues and University of Nantes, 44000 Nantes, France prospects for future generations of PPAR modulators. Biochim Biophys Acta
2007; 1771: 1065–81.
13 Nissen SE, Wolski K, Topol EJ. Eff ect of muraglitazar on death and major DC has received fees for consultancy, honoraria, speaking, travel, or adverse cardiovascular events in patients with type 2 diabetes. JAMA 2005; accommodation from Takeda, GlaxoSmithKline, Merck Sharpe & Dohme, 294: 2581–86.
AstraZeneca, Bristol Myers Squibb, Boehringer Ingelheim, Novo Nordisk, Roche, 14 Bénardeau A, Benz J, Binggeli A, et al. Aleglitazar, a new, potent, and balanced dual PPARα/γ agonist for the treatment of type II diabetes.
Bioorg Med Chem Lett 2009; 19: 2468–73.
Lincoff AM, Mudaliar S, Rabbia M, Chognot C, Herz M. Eff ect of 15 Gelman L, Neige J, Desvergne B. Molecular basis of selective PPAR the dual peroxisome proliferator-activated receptor-α/γ agonist aleglitazar modulation for the treatment of type 2 diabetes. Biochim Biophys Acta on risk of cardiovascular disease in patients with type 2 diabetes 2007; 1771: 1094–107.
(SYNCHRONY): a phase II, randomised, dose-ranging study. Lancet 2009; published online June 8. DOI:10.1016/S0140-6736(09)60870-9.
Reporting genetic association studies: the STREGA statement
The completion of sequencing of the human genome as quickly as expected.2 The devil is in the detail. For created high hopes for the translational potential instance, for the suggested relation between several of knowledge from genetic association studies. genetic polymorphisms and occurrence of sepsis in The combination of advances in genomic and human beings, systematic review in 2006 of the study related technologies and methods of conventional reports showed that the information needed to assess epidemiology has produced one of the most dynamic the risk of bias was lacking.3fi elds of medical research. Genome-wide association Results of medical research need to be reported with studies have led to an unparalleled growth and suffi cient accuracy and transparency to enable critical improved accuracy in genetic information.1 At the appraisal. Reporting guidelines aim to improve the same time, the complexity of the available data has research literature by the use of checklists with items increased substantially. However, advances in the deemed essential by multidisciplinary working groups.4 application of this knowledge, such as determination The STrengthening the Reporting of OBservational of an individual’s risk of disease or development of Studies in Epidemiology (STROBE) statement was personalised therapies, might not become available developed to help authors to improve the reporting www.thelancet.com Vol 374 July 11, 2009
Rationale for inclusion
Non-diff erential genotyping errors will usually bias associations towards the null; when there are systematic diff erences in genotyping according to outcome status (diff erential error) bias in any direction can occur This type of confounding can occur if study subpopulations diff er both in allele (or genotype) frequencies and disease risks, and if these subpopulations are unevenly distributed across exposure groups (or between cases and controls) In studies with a design covered by STREGA, haplotypes have to be inferred because of absence of available family information; because there are diverse methods for inferring haplotypes, they should be reported Departure from HWE might suggest errors or peculiarities in the data; reports of genetic associations do not always include information about conformity with HWE and some have limitations or errors in HWE assessment Publications that present and synthesise data from several studies in a single report are becoming more common; in particular, many genome-wide association analyses describe several study populations, sometimes with diff erent study designs and genotyping platforms, and in various stages of discovery and replication Selection bias can occur if genetic associations are investigated in one or more subsets of participants (subsamples) from a particular study, if there is diff erential non-participation in groups being compared, or if there are diff erential genotyping call rates in groups being compared Without an explicit rationale, potential for selective reporting of study results is diffi cult to judge; empirical evidence exists for other types of studies (eg, randomised trials) that the reporting of outcomes is frequently incomplete and biased in favour of statistically signifi cant fi ndings Study of a quantitative variable can be compromised when the trait is subjected to eff ects of treatment (eg, lipid-related trait for which several individuals are taking lipid-lowering drugs); without appropriate correction, this eff ect can lead to bias in eff ect estimates and loss of power Analysis methods should be transparent and replicable; genetic association studies are often done with specialised software, which should be reported Methods of analysis used in family-based studies are diff erent from those used in studies based on unrelated cases and controls; even in studies based on apparently unrelated individuals, some might be (distant) relatives; this possibility might need to be probed with appropriate methods and adjusted for in data analysis Synthesis of fi ndings across published studies depends on availability of suffi ciently detailed data from the individual studies Key problem is of possible selective reporting of false-positive results; type I errors are especially relevant in genome-wide association studies because large searches among genetic variants can be expected to fi nd positive results with odds ratios signifi cantly diff erent from 1 by chance alone STREGA=STrengthening the REporting of Genetic Association studies. HWE=Hardy-Weinberg equilibrium. Table: Rationale for areas covered by STREGA items that are specifi c to genetic association studies
of cross-sectional, case-control, and cohort studies.5 is neither prescriptive, nor does it pass judgment on Genetic association studies that use one of these study how researchers have done their studies. Rather, the designs present several specifi c challenges. As part statement advocates for transparency in reporting of its road map for human-genome epidemiology, how a study was done. This transparency might the Human Genome Epidemiology Network set up a help to inform the design, conduct, and analysis of working group to provide guidance on how to report future studies, and improve the reliable integration these studies.6 This initiative, called STrengthening the of evidence from multiple studies. STREGA can help REporting of Genetic Association studies (STREGA), authors to include the essential information in reports elaborated on the STROBE statement. The resulting of both smaller association studies and large-scale STREGA statement was recently published in several analyses of data from several sources. Earlier reporting journals simultaneously.7 The STREGA checklist guidelines, such as the CONSORT statement, have also provides additions to 12 of the 22 items of the proven useful for peer reviewers and editors.8 Evidence STROBE checklist to cover crucial aspects of genetic emerges that they also helped to improve the quality epidemiology reporting (table). Each of these areas of research reports.9 The endorsement by journals—eg, has a rationale given.7 in instructions for authors or reviewers—will be key In the fast-moving fi eld of genetic association to the success of the STREGA statement and of other studies, the risk of new methodological pitfalls is high. reporting guidelines. If, on top of enhanced accuracy The eff ects of individual variants in common disorders and transparency, the STREGA recommendations can are almost always modest. Generally, the credibility help to make articles on genetic association studies of gene–disease associations is low if the evidence an easier read for both specialised and less profi cient comes from single studies of small scale and cannot readers, it will have served its purpose.
be replicated. Genetic evidence can only transform the prevention, diagnosis, or therapy of diseases if *Erik von Elm, David Moher, Julian Little, on behalf of the it is consistent and strong. The STREGA statement STREGA collaboration www.thelancet.com Vol 374 July 11, 2009
German Cochrane Centre, Department of Medical Biometry and Clark MF, Baudouin SV. A systematic review of the quality of genetic Statistics, University Medical Centre Freiburg, D-79104 Freiburg, association studies in human sepsis. Intensive Care Med 2006; 32: 1706–12.
Altman DG, Simera I, Hoey J, Moher D, Schulz K. EQUATOR: reporting Germany (EvE); Swiss Paraplegic Research, Nottwil, Switzerland guidelines for health research. Lancet 2008; 371: 1149–50.
(EvE); Clinical Epidemiology Program, Methods Centre, Ottawa von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. Hospital Research Institute, Ottawa, ON, Canada (DM); and The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Department of Epidemiology and Community Medicine, Lancet 2007; 370: 1453–57.
University of Ottawa, Ottawa, ON, Canada (DM, JL) Ioannidis JP, Gwinn M, Little J, et al. A road map for effi human genome epidemiology. Nat Genet 2006; 38: 3–5.
Little J, Higgins JP, Ioannidis JP, et al. STrengthening the REporting of We thank France Gagnon, Julian Higgins, John Ioannidis, and Muin Khoury for Genetic Association studies (STREGA): an extension of the STROBE helpful comments. We are co-authors of the STREGA statement and are statement. Hum Genet 2009; 125: 131–51. http://www.medicine.uottawa.
involved in other reporting guideline initiatives, including CONSORT, STROBE, ca/public-health-genomics/web/assets/documents/Final- and the EQUATOR Network. Maja Zecevic, North American Senior Editor at STREGA%20manuscript-Feb2%202009.pdf (accessed June 22, 2009).
The Lancet, was part of the STREGA group.
Moher D, Simera I, Schulz KF, Hoey J, Altman DG. Helping editors, peer reviewers and authors improve the clarity, completeness and transparency McCarthy MI, Abecasis GR, Cardon LR, et al. Genome-wide association studies for complex traits: consensus, uncertainty and challenges. BMC Med 2008; 6: 13.
Nat Rev Genet 2008; 9: 356–69.
Plint AC, Moher D, Morrison A, et al. Does the CONSORT checklist improve the quality of reports of randomised controlled trials? A systematic review. Ioannidis JP. Personalized genetic prediction: too limited, too expensive, or
too soon? Ann Intern Med 2009; 150: 139–41.
Med J Aust 2006; 185: 263–67.
New WHO growth standards: roll-out needs more resources
See Comment page 94
On May 22, WHO and UNICEF issued a statement low weight-for-age.5 This defi nition was changed to endorsing new case defi nitions of severe acute weight-for-height6 to better identify children who malnutrition based on the 2006 WHO growth would benefi t most from treatment. Weight-for-standards.1 Before the global food crisis, a 2006 height expressed as Z scores is useful for surveys, yet review estimated that 13 million children had such many treatment programmes admit children with malnutrition.2 By January, 2009, aid agencies reported conceptually simpler %-of-median measures. (Minus about 19 million aff ected.3 Calls were made for the 1 Z score=1 standard deviation below a normally disease-burden demand to be met with increased supply distributed population median. Nutritional oedema of treatment services.3 Increases in the prevalence of is also part of the case defi nition for severe acute severe acute malnutrition associated with adoption of malnutrition.) More recently, focus has been on the new case defi nitions have been noted,4 but balances in treatment–service supply and demand are also crucial WHO growth standards are an international gold standard describing how children should grow when Numbers of children diagnosed as malnourished vary measured by weight and height.8 Previously, severe greatly depending on which case defi nition is used. acute malnutrition was defi ned as weight-for-height Each has advantages and disadvantages, and is useful <70% or <–3 Z scores below the National Centre for for particular purposes. Malnutrition for admission Health Statistics (NCHS) median. The new case defi nition to feeding programmes was originally defi ned by is weight-for-height <–3 Z scores below the WHO growth standards median. The diagnostic threshold of Number of cases
mid-upper-arm circumference has also been changed <70% of median weight-for-height (NCHS) from 110 mm to 115 mm. Increases in the diagnoses of <70% of median weight-for-height (NCHS), or severe acute malnutrition with the new WHO weight- <–3 weight-for-height Z scores (NCHS) for-height criteria have been noted.9–11 However, until <–3 weight-for-height Z scores (WHO) now there has been no documentation of which case Asia, 4; Eastern Africa, 8; Middle Africa, 1; Northern Africa, 2; Western Africa, 6; defi nition countries are using, and therefore the size of Southern Africa, 1. Guidelines include versions marked as fi nal or draft. All protocols classify oedematous malnutrition as severe acute malnutrition, and all accept the expected change in prevalence estimates for severe mid-upper-arm circumference <110 mm as defi ning such malnutrition.
Table: Case defi nitions of severe acute malnutrition in 22 national
In the table, we present treatment admission criteria guidelines
in 22 countries where severe acute malnutrition is www.thelancet.com Vol 374 July 11, 2009

Source: http://www.strobe-statement.org/fileadmin/Strobe/uploads/commentaries/von_Elm_Lancet_2009.pdf