Matching in case-control studies - A comment on, 'Do geriatrics require dose titration for antidiabetic agents?'

K Kaushal¹, SK Raina²,  
¹ Department of Community Medicine, Indira Gandhi Medical College, Shimla, India
² Department of Community Medicine, Dr. Rajendra Prasad Government Medical College, Tanda, Kangra, Himachal Pradesh, India

Correspondence Address:
K Kaushal
Department of Community Medicine, Indira Gandhi Medical College, Shimla
India

Full Text

Sir,

This has reference to the paper 'Do geriatrics require dose titration for antidiabetic agents?' published in J Postgrad Med (2013; 59:271-4). [1] The authors have conducted this prospective study for six months in a tertiary care hospital, where patients with type 2 diabetes mellitus were grouped into geriatric (age ≥ 60 years) and non-geriatric (age < 60 years) groups. The patients' demographic data, duration of diabetes, medication, and serum creatinine were recorded by the authors. [1]

We have concerns as to how the authors have defined the duration of diabetes on the basis of history alone. Knowing the duration of diabetes is important because a large number of individuals who meet the criteria for diabetes mellitus are asymptomatic and unaware of having the disease for up to a decade, before diagnosis. [2] Half of those with diabetes mellitus present with one or more diabetes-specific complications at the time of diagnosis. [2] We are also aware that the duration of diabetes may influence the complications of diabetes, including chronic kidney disease (CKD). Any further complication of diabetes is in itself a reason for dose titration.

Age is another important factor deciding dose titration. This was the aim of this study. However, duration of the disease may act as a potential confounder in this study. A look at [Table 1] used in this study explains the point further [Table 1]. [1] The table shows that there is a significant difference in the duration of disease between the two groups. The authors have attempted to use univariate analysis to adjust for the duration of disease and creatinine clearance (CrCl). But no details on the adjusted results have been provided in the study. {Table 1}

Matching is an answer to this. The authors could have matched the duration of diabetes and stratified it on other variables, including age. This would have identified age as a factor responsible for deciding the dosage for diabetes management. In a prospective study, matching must be done in advance. Using logistic regression analysis after the completion of a study, or univariate analysis (as used by the authors), retains the pairing, but necessitates adjustment for other variables that have not been matched.

References