Age and sex matching in case-control studies.

Sir,

Went through with interest article entitled “subclinical atherosclerotic vascular disease in chronic obstructive pulmonary disease: Prospective hospital-based case control study” published in Lung India (2015; 32: 137-41).[1] The present case-control study was designed to assess the relationship between sub-clinical atherosclerotic vascular diseases with chronic obstructive pulmonary disorder (COPD). The authors, at the end of the study conclude that the frequency of carotid plaqing is high in COPD patients.[1] Authors deserve appreciation for their effort. However, I have some concerns with the methodology adopted for the purpose of this study.

As per the authors, patients fulfilling the criteria for COPD were enrolled as cases and those who did not fulfil the standard diagnostic criteria were enrolled as controls.[1] The authors further state that there were 142 COPD patients and 124 age- and sex-matched controls without COPD and cardiovascular diseases.[1]

The study is a hospital-based case-control study and choosing suitable hospital controls is often difficult. After the cases and controls for a study have been determined, it is necessary to decide how many controls per case should be selected. When the number of available cases and controls is large and the cost of obtaining information from both groups is comparable, the optimal control-to case ratio is 1:1.[2] When the number of cases available for the study is small, or when the cost of obtaining information is greater for cases than controls, the control-to-case ratio can be altered to ensure that the study will be able to detect an effect, if one really exists (i.e., that the study has the necessary statistical power). The greater the number of controls per case, the greater the power of the study (for a given number of cases). However, there is generally little justification to increase this ratio beyond 4:1, because the gain in statistical power with each additional control beyond this point is of limited magnitude.[2] The authors in this study have used a smaller number of controls for cases which may not provide adequate power to the conclusion drawn from the study sample.

Further, a major disadvantage of a control group selected from diseased individuals (as done in this study) is that some of their illnesses may share risk factors with the outcome (carotid plaqing) under study, thereby a cause for bias. The authors seem to have inadequately handled this and may have left some room for bias to happen.

One way of minimizing this bias (as attempted by the authors) is matching for cases and controls. The authors claim that age- and sex-matched controls without COPD and cardiovascular diseases. But it does not seem that the authors have age and sex matched. If, cases and controls were individually matched on age and sex, then the number of cases and controls would have been equal and not unequal. Age and sex matching of controls means a similar proportion to the cases fall into the various categories defined by the matching variable (sex and age in this study). For instance, if 25% of the cases are males aged 65-75 years, 25% of the controls would be taken to have similar characteristics.