Incorrect statistical analysis: association among low whole blood viscosity, haematocrit, haemoglobin and diabetic retinopathy in subjects with type 2 diabetes.

Incorrect statistical analysis: association among low whole blood viscosity, haematocrit, haemoglobin and diabetic retinopathy in subjects with type 2 diabetes In the recently published study by Irace et al the authors describe selecting two groups of individuals, one with and one without type 2 diabetes, the latter matched to the former with respect to age and sex. Although not explicitly stated, the authors have employed a matched case-control study design, which is appropriate. They further describe using unpaired t tests and logistic regression for their statistical analysis, neither of which are appropriate. This is attributed to the fact that neither accounts for the matched nature of the study design. The authors are strongly urged to conduct a reanalysis of their study data and provide readers with the appropriate measures of association and p values, amending their interpretations as warranted.