Child Marriage in Rural Gujarat.

Sir,

I found the article on child marriage (CM) by Pandya et al. very interesting.[1] The authors successfully brought up the social issue to examine its effect on subsequent health issues. However, few points need consideration.

First of all, the p value used for assessing relation between socio economic status (SES)/education of parents and CM has been given too much importance. A close look at the table will reveal gradual change in proportion of CM with higher education and SES. In fact, if we combine illiterate and literate categories together in one group and rest of the categories in another group, we will find that education of husband's mother played a significant role in deciding child marriage (p = 0.006).[2]

Secondly, it is not clear how knowledge about family welfare methods dipped after marriage in non-CM group (as compared to knowledge at the time of marriage and current one) from 27 respondents to 11.

Thirdly, it was reported that “awareness regarding importance of early registration for ANC is less in women with CM”. It is difficult to conclude whether CM or illiteracy is responsible for late registration based on chi square analysis. Illiteracy and CM has been shown to be overlapping earlier. An approach through logistic regression would have provided more clarity in this issue.

Lastly, the authors might have indicated the analysis where they have used the Fisher's exact test. In the analysis of mother's age at the birth of first child, it seems that Fisher's test has been used here as a criteria for chi square; “No more than 20% of the cells should have expected frequencies of less than 5” is not applicable here.[3] For clarification, the expected values are given below.[4] (Table 1)

Table 1

Expected and observed values for assessing background characteristics

Financial support and sponsorship

Nil

Conflicts of interest

There are no conflicts of interest.

Sir,

We appreciate the keen interest shown by the author of this letter in our article. We present our views in response to the points raised by the author.

We analyzed the association of child marriage (CM) and education by categorizing the variable education in five appropriate categories. So it would not be logical to combine these categories in to two.

The respondents, though reported that they knew about family welfare methods at the time of marriage, could not recollect the same at the time of interview.

We agree with the author that logistic regression would have differentiated the effect of CM and education on awareness regarding importance of early registration for ANC. However, because of space constraints, it was not feasible to include

multivariate analysis in the article.

It is mentioned in ‘materials and methods’ section that statistical significance was tested using Chi2 test and Fisher's exact test. Here, 33% cells have expected values less than 5. Hence Fisher's exact test was applied.