PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of ijoemedHomeCurrent issueInstructionsSubmit article
 
Indian J Occup Environ Med. 2016 Sep-Dec; 20(3): 153.
PMCID: PMC5384395

Choosing controls in a case control study

Dear Sir,

This is in reference to the article entitled “Prevalence of musculoskeletal discomfort among the workers engaged in Jewellery manufacturing” published in Indian J Occup Environ Med (2015;19:44-55).[1] The author deserves credit for their effort. However, I have a few concerns with the design of this study. The author in the section “Subjects and Methods” states that cases selected for the purpose of this study were workers engaged in jewellery manufacturing (Group 1) with a minimum 1-year experience in the same job, whereas the controls selected for the purpose of this study were graduate students who were engaged in post-bachelor studies. The author further states that the controls were age-matched with the cases. Further, as an inference from the results, the author had chosen 110 controls against 500 cases.

My concerns primarily are with the way cases and controls have been chosen for this study. As a rule, the controls in a case-control study must fulfil all the eligibility criteria defined for the cases apart from those relating to the diagnosis of the disease. Further, the controls should represent the same population, from which the cases are drawn, i.e., they should provide an estimate of the exposure prevalence in the population from which the cases arise. If not, the results of the study are likely to be distorted because of selection bias. In effect, the controls should match cases except on the variable of interest. The control chosen by author for this study belongs to an entirely different occupational group (educational institutions), and therefore, represents a different population base.

This takes me to my other concerns. The author has chosen fewer controls (110) than cases (500). Choosing the number of controls is an important design decision, and is important to consider the ratio of controls to cases.[2] In an unmatched case-control design, there is usually little marginal increase in precision from increasing the ratio of controls to cases beyond four, except when the effect of exposure is large. However, the authors have chosen a ratio of 0.22 controls versus cases, which seems very inadequate. As a continuation of my concern, I am unable to understand how age matching between cases and controls was performed. This seems impossible with this ratio of controls: cases used in this study. Apparently, matching for age will require a ratio of 1:1 at the minimum.

Lastly, matching for just age alone would not have avoided all the confounding possible in this study as the amount and type of exposure between cases and controls is quite variable.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

REFERENCES

1. Salve UR. Prevalence of musculoskeletal discomfort among the workers engaged in Jewellery manufacturing. Indian J Occup Environ Med. 2015;19:44–55. [PMC free article] [PubMed]
2. Raina SK. Ratio of controls to cases: A design decision. J Postgrad Med. 2014;60:414. [PubMed]

Articles from Indian Journal of Occupational and Environmental Medicine are provided here courtesy of Wolters Kluwer -- Medknow Publications