Pharmacological and lifestyle interventions to prevent or delay type 2 diabetes in people with impaired glucose tolerance: systematic review and meta-analysis
We would like to thank Ingrid Mühlhauser for her comments. Ideally
all the important outcomes would have been considered in our meta-analyses
of diabetes prevention studies, unfortunately we were restricted by the
outcomes reported. As you point out in your paper(1) the DPP study did not
report results of HbA1c in their main publication, and neither blood
glucose nor HbA1c values were reported as outcomes in the core
publications of the STOP-NIDDM study, the smaller studies used in our meta
-analyses were further limited in their reported results.
It is true that taking a continuous variable and categorising it into
IGT and diabetes using arbitrary cut-offs is oversimplifying the problem,
especially when risk of complications increase with increasing blood
glucose and HBA1c levels. But the important issue for patients is whether
they have the disease or not. As only those classified as having diabetes
will receive treatment, it could be argued that presence or absence of
diabetes is the most important clinical outcome to report.
We agree that it is important to report results in a manner that
enables clinicians to make an accurate interpretation of the intervention
effect. To this end we reported our results not just as relative risks
but, as suggested in your paper (1), also in the form of absolute risk
reduction and numbers needed to treat.
1) Mühlhauser I, Kasper J, Meyer G, FEND. Understanding of diabetes
prevention studies: questionnaire survey of professionals in diabetes
care.
Competing interests:
Kamlesh Khunti has received sponsorship for attending conferences and small honoraria from pharmaceutical companies that manufacture hypoglycaemic and anti-obesity drugs.
Competing interests:
No competing interests
23 February 2007
Clare L Gillies
Medical Statistician
Keith R. Abrams, Paul C. Lambert, Nicola J. Cooper, Alex J. Sutton, Ron T Hsu, Kamlesh Khunti
Rapid Response:
Re: Glucose values should be reported
We would like to thank Ingrid Mühlhauser for her comments. Ideally
all the important outcomes would have been considered in our meta-analyses
of diabetes prevention studies, unfortunately we were restricted by the
outcomes reported. As you point out in your paper(1) the DPP study did not
report results of HbA1c in their main publication, and neither blood
glucose nor HbA1c values were reported as outcomes in the core
publications of the STOP-NIDDM study, the smaller studies used in our meta
-analyses were further limited in their reported results.
It is true that taking a continuous variable and categorising it into
IGT and diabetes using arbitrary cut-offs is oversimplifying the problem,
especially when risk of complications increase with increasing blood
glucose and HBA1c levels. But the important issue for patients is whether
they have the disease or not. As only those classified as having diabetes
will receive treatment, it could be argued that presence or absence of
diabetes is the most important clinical outcome to report.
We agree that it is important to report results in a manner that
enables clinicians to make an accurate interpretation of the intervention
effect. To this end we reported our results not just as relative risks
but, as suggested in your paper (1), also in the form of absolute risk
reduction and numbers needed to treat.
1) Mühlhauser I, Kasper J, Meyer G, FEND. Understanding of diabetes
prevention studies: questionnaire survey of professionals in diabetes
care.
Competing interests:
Kamlesh Khunti has received sponsorship for attending conferences and small honoraria from pharmaceutical companies that manufacture hypoglycaemic and anti-obesity drugs.
Competing interests: No competing interests