|LETTER TO THE EDITOR
|Year : 2011 | Volume
| Issue : 1 | Page : 53-54
Validity of results obtained from thirty clusters on "prevalence of iodine deficiency disorders" drawn from a large state in India
Umesh Kapil1, Padam Singh2
1 Professor, Public Health Nutrition, Human Nutrition Unit, All India Institute of Medical Sciences, New Delhi, India
2 Head-Research and Evaluation, EPOS Health India Pvt. Ltd., Gurgaon, Haryana, India
|Date of Web Publication||30-Jun-2011|
Professor, Public Health Nutrition, Human Nutrition Unit, Old OT Block, All India Institute of Medical Sciences, New Delhi - 110029
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Kapil U, Singh P. Validity of results obtained from thirty clusters on "prevalence of iodine deficiency disorders" drawn from a large state in India. Indian J Public Health 2011;55:53-4
|How to cite this URL:|
Kapil U, Singh P. Validity of results obtained from thirty clusters on "prevalence of iodine deficiency disorders" drawn from a large state in India. Indian J Public Health [serial online] 2011 [cited 2020 Jul 6];55:53-4. Available from: http://www.ijph.in/text.asp?2011/55/1/53/82560
We read the article entitled "A review of tracking progress towards elimination of iodine deficiency disorders in Tamil Nadu, India," in Indian Journal of Public Health, vol. 54, 3, July-September, 2010.  We found the manuscript extremely informative and useful. However, there are research methodological issue that needs discussion and clarifications. We would like to share these issues with the readers of this journal so that valid research methodology is followed in assessment of iodine deficiency disorders.
According to authors, a cross-sectional community-based study was conducted. A total of 1200 children were included from the 30 clusters selected from the entire state of Tamil Nadu which had a total population of 62.4 million.
The main issue that we would like to discuss here is whether 30 clusters (1200 children) selected from a population of 62.4 million could be representative of the iodine deficiency status amongst children, which is a proxy indicator of iodine deficiency in the entire population of Tamil Nadu?
The 30-cluster methodology was proposed for estimating the parameters at the district level. However, the same method is being followed for drawing inferences at the state level without understanding its limitations. It is important to bring on record that according to the theory of sampling, the findings from a sample can be generalized only if the sample adequately represents the cross-section of the target population. Thus, while drawing inference about the population universe based on a sample, the desirable requirements are
In order to ensure representativeness, the samples should ideally be selected using the stratified multistage design, in which following aspects of sampling have to be necessarily followed. 
- adequacy of sample to draw inference with desired precision and confidence level;
- representative character of the sample in order to ensure that the inferences drawn are valid for the universe.
- Stratification: there is inherent variability in the population that needs to be captured in the representative sample. To ensure this, the population has to be stratified into groups. The stratification has to be done in such a way that within strata variation is minimum, ie, strata are made as homogeneous as possible. Thereafter, sampling has to be done from each stratum. Importantly, the overall sample has to be allocated to different strata taking into account the size of the strata and the variability within strata.
- Multistage sampling: in multistage sampling, primary sampling units (PSUs) are selected by population proportionate to size (PPS) methodology, and within each PSU, equal numbers of secondary sampling units (SSU) are selected making design as self-weighting. This is to be done for each stratum. Further, in multistage sampling, the overall sample has to be allocated to different stages of sampling taking into account of the variability between PSUs and within PSUs, as well as contribution of each to the variance of the overall estimate. The most important consideration is to have the adequacy of the sample at all levels.
The surveys done by National Sample Survey Organization (NSSO) and National Family Health Survey (NFHS) make use of these statistical principals while designing their surveys. The results obtained through such surveys are reliable and robust, which are being used for program and policy purposes.
The "adequacy" of sample size and representativeness of the sample are two important considerations. Ideally, the number of clusters to be selected should constitute about 5% of the "total clusters" present in the population from which 30 clusters have been drawn. In any case, 30 clusters selected for survey should not be less than 1% of the total clusters present in the population to be surveyed. 
Further, we would like to mention is that the authors have calculated the sample size assuming the prevalence of adequately iodized salt samples in Tamil Nadu. However, they have primarily concluded on the estimation of prevalence of IDD, which was clearly not the criterion for estimating the sample size. The design effect is never expressed in "percentages," as it is an absolute number. 
Another issue that merits to be mentioned is that in this manuscript authors have made wrong statement on "conflict of interest." This research paper was financially supported by Micronutrient Initiative. One of authors from this issue is from Micronutrient Initiative. Hence, there is a conflict of interest in the results communicated through this manuscript.
| References|| |
|1.||Pandav CS, Krishnamurthy P, Sankar R, Yadav K, Palanivel C, Karmarkar MG. A review of tracking progress towards elimination of Iodine deficiency disorders in Tamil Nadu, India. Indian J Public Health 2010;54:120-5. |
|2.||Sukhatme PV, Sukhatme BV. Sampling theory of surveys with applications. 3rd ed. Ames, Iowa, U.S.A. and New Delhi, India: Iowa State University Press; 1984. |
|3.||William G. Sampling Techniques. 3rd ed. Cochran: Wiley; 1977. p. 109-20. |
|4.||Joint WHO/UNICEF/ICCIDD Consultation. Indicators for assessing Iodine Deficiency Disorders and Their Control Programmes." Geneva: WHO; 1992. p. 78-97. |