Conducting case control study on a cross-sectional database.

Sir,

This is in reference to an article entitled “pattern and risk factors in the young among stroke patients admitted in medical college hospital, Thiruvananthapuram” published in Ann Indian Acad Neurol 2015;18:20-3.[1] The authors need to be appreciated for their effort in planning a study on risk factors for stroke among young patients. However, I have a few concerns with the way this study has been presented. The authors state that the study was conducted as a cross-sectional study with case-control comparison, but the case-control part of the study has not been dealt with adequately. Only 15 cases have been chosen for comparison with 85 controls. A small sample does not return us with a good odds ratio and reduces the power of the study. I understand the limitation of picking up cases and controls from a cross-section study base. The limitation could have been removed by conducting a matched case-control study rather than the above unmatched design.

In Material and Methods section of the article under heading “Controls”, the authors state that in case, if the patient was not able to give answers to the questions, necessary information was obtained from the bystander and other available sources. A bystander by definition “is a person who, although present at some event, does not take part in it”.[2] In other words a bystander is an observer or spectator. How can a bystander provide necessary information of the patient? I think the authors need to have a relook on this statement.

In the Results section under heading “Risk factors”, a comparison of frequencies of stroke risk factors among cases and controls has been provided and various risk factors have been enlisted (Table 2 of the study).[1] However, I am not sure as to how patients with comorbid conditions like diabetes and hypertension together, were placed in this list. Was it that all patients included in this study suffered from a single condition only? Did physical inactivity, smoking, and alcohol exist in isolation? A look at the numbers suggests otherwise. So how did the authors choose individual risk factors? The pattern has been repeated in Tables 3 and 4 of the study[1].