Health Habits of Children

416 Brief communication HEALTH HABITS OF CHILDREN To the editor: I read the article by LJ. Cornelius, Ph.D. [Vol. 2, No. 3:374395 ] with a great deal of expectation; we have replicated the Alameda health practices and indexes1-2 in adults, and I was interested in discovering if similar associations occurred in children. I must say that my curiosity was only partially satisfied. Dr. Cornelius used secondary analysis of data collected in the National Medical Expenditure Survey (NMES) to study the association between demographic and socioeconomic variables and four health habits [eating breakfast, maintaining desirable weight, sleeping regularly, and wearing seat belts] in a subsample of 6,722 children between the ages of five and 17. The independent variables included gender, age, ethnicity/race, education, poverty level (these last two measured at household, or head of household, level), and size of residence. After an interesting introduction to health habits in children, he proceeds to show simple bivariate associations. These are more or less predictable and should be considered as exploratory. For almost all of the health habits, ethnic/racial background mediates the associations with the other variables. A correlation matrix showed very small association between health practices. (I must point out that the author apparently used Pearson's correlation, but Spearman's was indicated, given the level of measurement of the variables. Nevertheless, and due to the large sample size, it is very likely that the value of the coefficients would be similar.) A multivariate analysis, although confirming the results of the bivariate analysis, was not satisfactory, due to the extremely small R2 (.0183). Only two variables—education and poverty level—had significant coefficients; again, the relatively large sample size would produce statistically significant results with small actual associations. But my main disappointment is related to the absence of an analysis between the health practices and health status. I am sure that the NMES survey instrument included a question about self-reported health status, and it would have been extremely interesting to know if the well-known association between health status and health practices, individually or in simply additive indexes3, existed in children. Another issue of potential interest would have been to study possible associations between health status and health practices, separately by Journal of Health Care for the Poor and Underserved, Vol. 2, No. 4, Spring 1992 ___________________________________________________________417 the two age groups considered, speculating that children aged five to 12 are more likely than teenagers to follow parental rules. Another issue which is not discussed is why important practices such as smoking, drinking, and exercise were omitted. The fact that proxy respondents —mostly mothers—were used may have affected the validity of reported smoking and drinking behaviours, even in teenagers, but exercise should not be affected by this, and it would have added an important practice to the analysis. In summary, this is an interesting paper on a subject of great importance. It has whetted our appetite for further and more elaborate analysis of the same database. Jorge Segovia, M.D., M.P.H. Professor of Social Medicine and Associate Dean of Community Medicine Faculty of Medicine Memorial University of Newfoundland HSC 2837 St. John's, Newfoundland Canada AlB 3V6 REFERENCES 1. Segovia J, Bartlett R, Edwards A. The association between self-assessed health status and individual health practices. Can J Public Health 1989 Jan/Feb;80:32-7. 2. Segovia J, Bartlett R, Edwards A. Health status and health practices—Alameda and beyond. Int J Epidem 199120(1 ):259-63. 3. Berkman L, Breslow L. Health and ways of living—The Alameda County Study. New York: Oxford University Press, 1983. Dr. Cornelius responds: While it is encouraging to hear from a colleague who has an interest in this area, I fear that Dr. Segovia's comments may reveal someconfusion about the study and about the nature of the data. First, this was not a secondary analysis of another agency's data. We at AHCPR have purposelycollected data on health status, use, and expenditures so that analysts can examine questions such as the ones posed in this paper. This brings me to my second point. Given the apparent vacuum of information in this area, even seemingly "simple bivariate...