Sir,
We read with interest the work by Sonali Jindal et al.,Citation1 which reports that platelet distribution width (PDW) was higher in diabetic patients with complications as compared to those without. Only two input variables (PDW and mean platelet volume (MPV)) qualified for the final discriminant model by stepwise discriminant analysis using age, duration of diabetes, platelet count, and platelet indices. They concluded that discriminant analysis using PDW and MPV could classify majority of patients with diabetic complications. In our opinion, this conclusion lacks enough proofs because of potential analytic and methodological shortcomings. It would, however, be appropriate if some further analyses were performed before such a conclusion is advanced.
Firstly, the duration of diabetes (subjects without complications: (56.5 ± 79.12) months and subjects with complications: (82.26 ± 78.53) months) were compared using the Mann–Whitney test. The standard deviation is close to the mean. The skewed data should be expressed using median instead of mean ± standard deviation.
Secondly, the use of discriminant functions without regard to methodology can result in erroneous answers. They can permit random assignment of each subject to a training set (85% of the subjects, used to create the discriminant) and a test set (15% of the subjects, used to estimate prospective classification). A more reliable estimate of prospective classification success results from a ‘leaving 15% out’ test. Generally, the discriminant function has practical value when its misjudgment rate is less than 0.1 or 0.2.Citation2
Thirdly, gender, lipid profile, HbA1c, hyperglycemia, hypertension, and the metabolic syndrome are also risk factors of microvascular complications in type 2 diabetes patients.Citation3–Citation5 Some of them should be collected and taken as input variables in stepwise discriminant analysis. If noncollinear variables were included in the analysis, there might be more confidence in the indicator of PDW as being related to microvascular complications in diabetic patients. Then, if the authors still find PDW as a predictor in the multivariate model(s), their conclusion is valid.
Editor's comment
The authors of the original paperCitation1 were invited to respond to this letter but have not yet done so.
References
- Jindal S, Gupta S, Gupta R, Kakkar A, Singh HV, Gupta K, et al. Platelet indices in diabetes mellitus: indicators of diabetic microvascular complications. Hematology. 2011;16:86–9.
- Xia JL. Discriminant analysis. In: , Sun ZQ, Xu YY (eds.) Medical statistics. 3rd ed. Beijing: People's Medical Publishing House; 2010. p. 326–38.
- Liu F, Bao Y, Hu R, Zhang X, Li H, Zhu D, et al. Screening and prevalence of peripheral neuropathy in type 2 diabetic outpatients: a randomized multicentre survey in 12 city hospitals of China. Diabetes Metab Res Rev. 2010;26:481–9.
- Reasner CA. Reducing cardiovascular complications of type 2 diabetes by targeting multiple risk factors. J Cardiovasc Pharmacol. 2008;52:136–44.
- Abdul-Ghani M, Nawaf G, Nawaf F, Itzhak B, Minuchin O, Vardi P. Increased prevalence of microvascular complications in type 2 diabetes patients with the metabolic syndrome. Isr Med Assoc J. 2006;8:378–82.