"의학 연구자료 분석과 통계적 기법"에 대하여

송혜양 한국통계학회 1993 응용통계연구 Vol.6 No.2

이 논문은 크게 두 부분으로 구성되었는데, 첫째 부분은 의학연구 목적에 따라 요구되는 분석법을 소개하고 있고, 둘째 부분은 국내 의학논문에서 통계처리의 잘못의 유형과 빈도를 제시하고 있습니다. 첫째 부분인 여러 분석법의 소개에서는 분석법의 열거로 보아서는 통계인들에게 새로운 것이 많지 않다고 언뜻 느껴지시겠지만, 실제로는 여러 면에서 새로운 점이 많습니다. 이 차이점을 살펴보면 우선 rate와 proportion과 ratio란 용어를 구분하여 사용하고 있고, 이들의 사용이 요구되는 의학적인 경우도 분명히 다름을 의학연구의 예제와 연구목적에 따라 제시하였습니다. 교과서처럼 쓸 수 있어서 설명이 부족합니다만, 의학에서 요구되는 구체적인 상황은 매우 다릅니다. 여러 가지 연구의 하나인 원인성연구를 살펴보면 measure of association으로써 rate difference (RD), rate ratio (PR)과 odds ratio (OR)의 세가지가 제시되었고, 이 measure에 대해서도 각각이 요구되는 의학적인 경우가 분명히 다르며, 또한 각각에 따른 분석기법도 다름을 논문에서 볼 수 있습니다.

兒童行動目録(CBCL) 에 의한 國民學生의 行動問題 調查 —兒童의 行動問題 研究( III ) ᅳ

홍강의,송혜양,김중술,홍경자,박성수 대한신경정신의학회 1988 신경정신의학 Vol.27 No.2

This is an epidemiological study of emotional/behavior problems in Korean elementary school children (total 2,399 • 1,236 boys, 1,163 girls) by empolying Children’ s Behavior Check List (CBCL) developed by Achenbach and Edelbrock(1983), to which 3 problem items were added for this study. A test-retest reliability of CBCL was was assessed by computing the Spearman- Brown correlation and the average correlation among items was. 72. The interparent agreeement was also assessed and the average correlation between mother and father was. 62. The study subjects were drawn by stratified sampling from 5 schools in Korea ; 3 schools (A, B, C) from Seoul, one school(D) from a medium size city and one school (E ) from a rural town. School C is located in upper/middle-upper class residential area, School A and D in middle class, and School B and E in lower-middle/lower class. The present paper reports the prevalence of behavioral/emotional problems in relation to sex, grade, locality and socioeconomic variables. Some of the major findings were as follows * 1) The mean total problem score excluding 3 additional items was 27.28 and did not differ significanthy according to sex and locality (Seoul vs non-Seoul), However, the mean total problem scores were significanthy different from school to school. The lowest score was observed in children of School C(22.15), higher scores for School B (30.89) and E (30.64) and in-between scores for the School A(27.58) and D (25.47). Father’ s education also had significant effect on the total problem scores • 31.46 for elementary or less, 27.79 for middle and high school and 23.05 for college or more. As for to the grade, mean total score was lowest for the first graders(22.12),increasing up until 4th grade(29.83), then decreasing for 5th graders(26.81) and then relatively high score for the 6th graders(29.81). 2) Examination of each 122 problem items revealed that prevalence rate ranged from 2.1% to 61.7% : less than 5% for 13 items, 5—10% for 18 items, 10~20% for 31 items, 20 ᅳ 30% for 25 items, 30—40% for 15 items, 40—50% for 13 items, more than 50% for 7 items.Although sex did not make any difference in total problem scores, examination of the prevalence of each item revealed that 47 problem items were significanthy different between sexes : 26 items were more prevalent in boys and 21 items in girls. The items more prevalent for boys were externalizing problems such as hyperkinetic, aggressive and other behaviors and 21 items, more prevalent for girls were internalizing problems such as emotional and psychosomatic problems. School grade had significant effect in 47 items, only 3 of which showed lower prevalence in the younger children and the prevalence of 39 items were higher in older children. Significant differences were observed in 83 items, and 94 items, respectively according to school and father* s education that could be considered to reflect socioeconomic status and it was reavealed that the lower the status, the higher the prevalence in most items. Items more common in upper class tend to be more anxiety-tension related( internalizing) and items more common in lower class tend to be more aggressive, hyperactive and other behavioval problems(externalizing). 3) Brief cross-cultural comparisons between the current study and other studies reported in the literature revealed major differences in several areas. The total problem score for Korean children was much higher than those of American and Dutch children, but lower than those of Canada and Chile. Although most of other studies reported higher total problem scores for boys, than for girls there was no sex difference in total problem score for Korean children. This was reflected in the finding of an almost equal numbers of items(26 for boys, 21 for girls) being more common than the other sex.

兒童行動目錄(CBCL)에 의한 國民學生의 行動問題 調査

홍강의,송혜양,김중술,홍경자,박성수 大韓神經精神醫學會 1988 신경정신의학 Vol.27 No.2

This is an epidemiological study of emotional/behavior problems in Korean elementary school children(total 2,399 : 1.236 boys, 1,163 girls) by empolying Children's Behavior Check List(CBCL) developed by Achenbach and Edelbrock(1983), to which 3 problem items were added for this study. A test-retest reliability of CBCL was was assessed by computing the Spearman-Browm correlation and the average correlation among items was. 72. The interparent agreeement was also assessed and the average correaltion between mother and father was. 62. The study subjects were drawm by stratified sampling from 5 schools in Korea ; 3 schools (A, B, C) from Seoul, one school(D) from a medium size city one school (E) from a rural town. School C is located in upper/middle-upper class residential area, School A and D in middle class, and School B and E in lower-middle/lower class. The present paper reports the prevalence of behavioral/emotional problems in relation to sex, grade, locality and socioeconomic variables. Some of the major findings were as follows ; 1) The mean total problem score excluding 3 additional items was 27.28 and did not differ significanthy according to sex and locality(Seoul vs non-Seoul), However, the mean total problem scores were significanthy different from school to school. The lowest score was observed in children of School C(22.15), higher scores for School B(30.89) and E(30.64) and in-between scores for the School A(27.58) and D(25.47). Father's education also had significant effect on the total problem scores : 31.46 for elementary of less 27.79 for middle and high school and 23.05 for college or more. As for to the grade, mean total score was lowest for the first graders(22.12), increasing up until 4th grade(29.83), then decreasing for 5th graders(26.81) and then relatively high score for the 6th graders(29.81). 2) Examination of each 122 problem items revealed that prevalence rate ranged from 2.1% to 61.7% : less than 5% for 13% items, 5-10% for 18 items, 10-20% for 31 items, 20-30% for 25 items, 30-40% for 15 items, 40-50% for 13 items, more than 50% for 7 items. Although sex did not make any difference in total problem scores, examination of the prevalence of each item revealed that 47 problem items were significanthy different between sexes : 26 items were more prevalent in boys and 21 items in girls. The items more prevalent for boys were externalizing problems such as hyperkinetic, aggressive and other behaviors and 21 items, more prevalent for girls were internalizing problems such as emotional and psychosomatic problems. School grade had significant effect in 47 items, only 3 of which showed lower prevalence in the younger children and the prevalence of 39 items were higher in older children. Significant differences were observed in 83 items, and 94 items, respectively according to school and father's education that could be considered to reflect socioeconomic status and it was reavealed that the lower the status, the higher the prevalence in most items. Items more common in upper class tend to be more anxiety-tension related(internalizing) and items more common in lower class tend to be more aggressive, hyperactive and other behavioval problems(externalizing). 3) Brief cross-cultural comparisons between the current study and other studies reported in the literature revealed major differences in several areas. The total problem score for Korean children was much higher than those of American and Dutch children, but lower than those of Canada and Chile. Although most of other studies reported higher total problem scores for boys, than for girls there was no sex difference in total problem score for Korean children. This was reflected in the finding of an almost equal numbers of items(26 for boys, 21 for girls) being more common than the other sex. Analyses by grade(age) also indicate major differences between Korean and American-Dutch children ; In contrast to the finding that total problem score was higher and larger number of items were more common for younger children in American and Dutch studies, they were higher for older children in Korean study. Socioeconomic variables seems to exert far more significant effect on the total score and the prevalence of each items in Korean study, when compared with American and Dutch studies. Further analyses of the data including a comparison with a clinic sample are deferred to another paper and so are the factor analysis results of 122 items into emotional/behavioral syndromes.

유방암 환자에서 술후 타목시펜요법의 의의

오세정(Se Jeong Oh),정상설(Sang Seol Chung),박우찬(Woo Chan Park),조원일(Won Il Cho),김정수(Jeong Soo Kim),최혜승(Seung Hye Choi),송혜양(Hae Hiang Song) 대한외과학회 1999 Annals of Surgical Treatment and Research(ASRT) Vol.57 No.3