motor skills than the Denver normative group. 4. How do you justify the high correlation of your DDST scores by pooling all ages together? There is such low correlation by age ranges in your Table 2. Evaluative (normative and correlative) studies of developmental screening tests are needed. All the information that you can provide to assist readers of your study would be appreciated. Helene S. Thorpe, MD Pediatric Consultant Assistant Clinical Professor Department of Pediatrics UC Davis School of Medicine Dav,is, CA

BARNES RESPONDS Dear Dr. Thorpe: Thank you for your letter of May 22, 1975, and for raising the questions that you did; I shall attempt to answer them in the order presented. 1. The basis for the numerical scores in Figures 1 to 4 and Table 3 was 1 point for each total item passed. The means and standard deviations were then calculated accordingly for each age category and subtest area. The earlier items were scored as "passed." This procedure was followed because in random spot checks carried out on the sample tested, some chilTABLE 1-Comparison of Age Distribution for Barnes and Stark versus Frankenberg and Dodds Normative Samples Frankenberg Barnes Age Groups

& Stark

No.

%

27-45 46-78

28 28 25 26 25 27 31 36

12.39 12.39 11.06 11.50 11.06 11.95 13.72

Total

226

months 1-3 4-6 7-9

10-12 13-16

17-26

& Dodds

No.

%

15.93

117 121 119 116 124 132 130 177

11.29 11.68 11.49 11.20 11.97 12.74 12.55 17.08

100.00

1036

100.00

dren were asked all of the questions up to their chronological age levels and no items were failed at an earlier level. The fact that the scores for age group 46 to 78 months did not reach one standard deviation above the mean may reflect a lack of sufficient items for screening children approximately 4 years of age and older; it may also reflect the possibility that the items currently asked are of such a similar difficulty level that they do not adequately discriminate across children 46 to 78 months of age. 2. As we were attempting to replicate the original DDST normative study as closely as possible, the distribution of our sample by age groups was as similar to the Frankenberg and Dodds (1967) distribution as we could make it. Table 1 compares the two distributions. The rationale given by Frankenberg and Dodds (1967) for heavily weighting their sample below age 2 was: "A greater number of children were tested at younger ages, when developmental changes occur faster." 3. I really don't have an answer to your question on whether child-rearing patterns in British Columbia are more similar to the American than to the British-European pattern. Although at first glance one may well favor the hypothesis of similarity to the American pattem because of our geographic proximity to the United States, the figures reported in the last Canadian Census (1971) indicate that 95 per cent of the population of the South Okanagan are either British or European or of British-European descent. Actually, because the Bryant, Davies, and Newcombe (1974) study does not report means and standard deviations by age groups for their normative sample, I really don't know whether Welsh children have significantly earlier language and fine motor skills than children born in the United States. I note that in their article the authors (page 479) do not claim that the differences they found were statistically significant. In fact, when one looks at the Denver and Cardiff norms one notes that there is considerable variation in scores, and I suspect, because of the respective range of scores for the four subtest areas across the two normative

samples, that the differences reported by Bryant et al. are probably not statistically significant. 4. Your question on the high correlation between subtest areas when the sample as a whole is looked at as compared to the separate age categories is one which I think can best be answered in terms of degree of heterogeneity of the sample. Statistically, the magnitude of a correlation coefficient usually varies significantly with the degree of heterogeneity of the sample being studied. In our case we might expect to find relatively low correlations when looking at our sample by age ranges, because both variables, age and subtest scores, have very restricted ranges (low heterogeneity); however, when we look at our sample as a whole (all ages combined), we do not have such range restrictions and thus find substantially higher correlations. Once again thank you for raising the questions that you did; I hope I have at least answered some of them satisfactorily. Keith E. Barnes, PhD South Okanagan Mental Health Center Kelowna, British Columbia Canada

ON INFANT MORTALITY In a recent paper Naomi Morris et al.I estimated that 27 per cent of the reduction in the U.S. infant mortality rate between 1965 and 1972 could be attributed to changes in the maternal age and birth order distribution of births. This estimate was based on a calculation of expected infant deaths derived by applying the age-birth order-specific infant mortality rates of 1960 to the age-birth order distribution of births between 1965 and 1970. It was assumed that the age-birth orderspecific infant mortality rates of 1960 remained constant over the period of observation; but this assumption may not be justified. Studies in Britain have shown that the age- and parityspecific rates for neonatal and postneonatal mortality declined significantly over the 15-year period between 1949 and 1964. Neonatal mortality LETTERS TO THE EDITOR 1113

risks declined by 26 to 34 per cent and postneonatal mortality risks declined by 52 to 58 per cent for different maternal age and parity categories.2 Though more recent data are not available for the U.K., it is probable that this trend has persisted to some degree. It is difficult to find U.S. studies which provide comparability over time, but there is some evidence of a decline in the risk of death for certain birth orders.3 If the age-birh orderspecific infant mortality risk has declined since 1960, Morris' calculations would overestimate the expected deaths and exaggerate the apparent impact of changing fertility on infant mortality. The paper by Morris et al. also failed to consider infant mortality trends by race. The infant mortality for the U.S. white population declined by 8 per cent between 1965 and 1969, whereas infant mortality for all other races declined by 20.2 per cent over the same period.5 Thus the decline in national infant mortality between 1965 and 1972 could in part be related to changes in differential mortality between races, which may have no bearing upon changes in fertility. It is noteworthy that infant mortality declined by 4.2 per cent in the U.K. and 20 per cent in the U.S. over the period 1965 to 1970, despite comparable changes in fertility. This would suggest that factors other than the change in the age-birth order distribution were largely responsible for the improvements in infant mortality in the United States. A reduction of births to mothers in high risk age and parity categories will undoubtedly have a beneficial effect on infant mortality. However, one cannot accept the magnitude of the estimates cited by Morris et al, in view of the potential biases and the importance of disturbing variables such as race. R. H. Gray, MD London School of Hygiene and Tropical Medicine REFERENCES 1. Morris, N. M., Udry, J. R., and Chase, C. L. Shifting Age-Parity Distribution of Births and the Decrease in Infant Mortality. Am. J. Public Health 65:359-362, 1975. 1114 AJPH OCTOBER, 1975, Vol. 65, No. 10

2. Spicer, C. C., and Lipworth, L. Regional and Social Factors in Infant Mortality. Studies on Medical and Population Subjects, No. 19, pp. 15-17. Her Majesty's Stationery Office, London, 1966. 3. MacMahon, B., Kovar, M. G., and Feldman, J. J. Infant Mortality Rates: Relationship with Mother's Reproductive History. United States. Vital Health Stat. Series 22, No. 15, p. 7, 1973. 4. Gendell, M., and Hellegers, A. E. The Influence of the Changes in Maternal Age, Birth Order and Color on the Changing Perinatal Mortality, Baltimore. 1961-66. Health Serv. Rep. 88:733-742, 1973. 5. Mortality Trend: Age, Color and Sex. United States, 1950-69. Vital Health Stat. Series 20, No. 15, 1973.

SEIDEN QUERIED Dear Sir: The questionnaire in "Patterns of Marijuana Use among Public Health Students" (AJPH 65:613-621, 1975) was particularly interesting in that it never comes right out and asks "Do you now use marijuana?" Presumably, Table 2 and the use pattern discussion on page 614 were derived from analyzing various combinations of responses to questions 7 and 8. If so, a possibly important point was not mentionedthat no one who had used marijuana more than "a few times" had discontinued usage. That is the only analysis compatible with the text. Regarding other drugs, the authors conclude on page 618 that it did not "appear that marijuana use necessarily led to harder drugs" and that "there was surprisingly little carryover to the use of other drugs." The only pertinent question on the questionnaire is number 11, which refers only to "ever" using other drugs. How do the authors put a time frame on such usage as a basis for stating that other drug usage occurred before, during, or after marijuana use? The authors state on page 614 that "use of 'harder' drugs was substantially lower" than marijuana use. This is confusing, since combined usage of these other drugs totals 97 per cent as contrasted to 76 per cent for marijuana (not corrected for multidrug usage). Would it be more precise to say that "use of any single harder drug was substantially lower than marijuana use, but use of at least one harder drug

was--"? It is not possible to tell this from the text. Authors' comments would be appreciated. My interest is in questionnaire analysis, not the pros and cons of

marijuana. John L. S. Hickey, MSPH Chapel Hill, NC

SEIDEN RESPONSE Dear Mr. Hickey: I am very flattered that you took the time to read our paper so thoroughly and I hope to clear up the problems of interpretation that you have experienced. William of Occam proclaimed that one should not multiply entities beyond necessity. This includes questionnaire items. It was redundant for us to ask whether a person used marijuana, since this could be ascertained from items 7, 8, and 10. It is true, as you point out, that no one who had used marijuana more than a few times had discontinued usage. Apparently the experimentation and rejection takes place early on and by the time someone is using regularly he is apt to maintain this schedule. As regards the relationship of marijuana use and use of other drugs, you are quite correct that there is no necessary time sequence. We were guided by the typical presumption that the use of marijuana leads to harder drugs-an axiom which has been advanced by law enforcement officials for a number of years. On the other hand, I suppose it is possible, if unlikely, that a person could progress from opiates to marijuana. To be more accurate, we should have stated that marijuana use was not necessarily related to the use of harder drugs. As to the last point, the logic of your argument is completely perplexing to me. To the best of my understanding you have summed the different responses to question 11 (p. 621). However, to do this one would have to assume that the responses were mutually exclusive, which they are not. In fact multiple drug usage is quite the rule and two of the respondents accounted for almost all the use of the harder and more exotic drugs, e.g., opiates, kava, and psilocybin. Accord-

Letter: On infant mortality.

motor skills than the Denver normative group. 4. How do you justify the high correlation of your DDST scores by pooling all ages together? There is su...
419KB Sizes 0 Downloads 0 Views