Health Behavior Changes in the United States, the United Kingdom, and France SHELDONM. RETCHIN, MD, MSPH, JAMESA. WELLS, PhD, ALAIN-JAC(,)UES VALLERON, DFSc, GARYL. ALBRECHT, PhD Objective: To determine changes in health behaviors in the United States, the United Kingdom, a n d France over the previous two years. Design: Cross-sectional survey o f nationally representative samples. Setting/participants: Surveys conducted betweenJune a n d November 1988 on p e r s o n s aged 16 to 50 y e a r s in the United States ( n = 1,940), the United Kingdom ( n = 1,833), a n d France ( n = 2,294) regarding health behaviors, attitudes toward health, a n d changes in health practices during the previous two years. Measurements and m a i n results: Using Bonferroni's adj u s t m e n t f o r multiple comparisons, residents o f the United States h a d significantly ( p < 0.05) higher Quetelet indices a n d reported higher egg a n d red meat consumption, bat had lower alcohol consumption, than did residents o f either the United Kingdom o r France. Americans were also significantly more likely to report attitudes accepting personal responsibility f o r their health a n d much m o r e often endorsed the role o f health behaviors (e.g., e x e r c i s e ) f o r decreasing the risk o f cardiovascular disease. Changes in health behavior over two y e a r s were consistently more likely in the United States f o r weight loss, decreased alcohol consumption, decreased red meat a n d egg consumption, a n d increased exercise. Americans were also much more likely to have changed at least three health behaviors in the previous two years (United States 41.5%, United Kingdom 25.5%, France 13.8%, p < 0.002). A multivariate linear model confirmed the high likelihood o f health behavior changes in the United States compared with the United Kingdom o r France. Conclusions: Thef i n d i n g s confirm that changes in health behaviors are continuing to occur in the United States, but remain comparatively modest in the United Kingdom a n d France. These international variations in health behaviors parallel differential declines in mortality rates in ischemic h e a r t disease. Key words: health behaviors; lifestyle; international health; risk factors; heart disease. J GEN INTvacN MED
1992;7:615 -622. NUMEROUS EPIDEMIOLOGICSTUDIESover the last several decades have demonstrated the relationship of selected health behaviors to physical health status and disease.t. 2 Since that time, major health p r o m o t i o n campaigns have been introduced, and modification of Received from the Project HOPE Center for Health Affairs, ChevyChase, Maryland (SMR,JAW),the Department of Internal Medicine, Medical College of Virginia, Virginia CommonwealthUniversity, Richmond, Virginia (SMR), the Institut de la Sant~ et de la Recherche M~dicale, Paris, France (A-JV), and the School of Public Health, Universityof Illinois-Chicago, Chicago, Illinois (GLA). Presented in part at the annual meeting of the Societyof General Internal Medicine, May 1- 3, 1991, Seattle, Washington. Address correspondence and reprint requests to Dr. Retchin: Box 287, MCVStation, Richmond, VA23298.
health behaviors has b e c o m e a major health care focus in the United States. 3 Distinguished authorities, such as the Institute of Medicine, have endorsed the role of health behaviors in the cause o f specific medical illnesses, particularly ischemic heart disease. 4 Finally, considerable media attention has been concentrated on health and illness, and as a result of these and other efforts, there have been substantial health behavior changes, s Although there is evidence that health behaviors have c h a n g e d substantially in recent years, less is k n o w n about international trends and comparisons a m o n g countries. International disparities in health behaviors are important to examine for several reasons. First, differences in health behaviors b e t w e e n countries are important because they may provide insights regarding changing patterns of disease. For example, although meaningful declines in ischemic heart disease have been observed over the last decade and a half in Northern and Western Europe ( 1 0 - 14%), by far the largest reductions have taken place in the United States -> 25%). 6 If health behaviors are at least partially responsible for observed variations a m o n g countries in rates of decline, then differences in corresponding health behaviors w o u l d be e x p e c t e d to parallel these variations. Second, changes in health behaviors in different societies may be the result of distinguishing sociocultural factors, 7, 8 and these factors are often difficult to investigate within countries. Since there are notable cultural influences that affect these factors, studies that involve international comparisons are important. As an example, following p r o f o u n d secular changes in the United States during the '50s and '60s, the p r o p o r t i o n o f w o m e n w h o s m o k e d rose considerably. As a result, rates of lung cancer in w o m e n are rising in the United States disproportionate to rates in some other countries. Last, the dissemination of scientific evidence regarding health p r o m o t i o n is strongly affected by the medical c o m m u n i t y , p u b l i c receptivity, and other factors that are peculiar to each society. Media interest and exposure for p u b l i c health issues in different countries are d e t e r m i n e d by p o p u l a r concerns, cultural environments, and g o v e r n m e n t regulations. Because these cultural characteristics are difficult to examine within countries, international differences in health attitudes may provide important opportunities for examining o u t c o m e s as a result of the different exposures to these influences. As part of an international study of behavior 615
616
Retchin et aL, HEALTHBEHAVIORCHANGES
changes related to the transmission of AIDS and cardiovascular disease, w e c o n d u c t e d a survey in the United States, the United Kingdom, and France. This survey i n c l u d e d questions regarding health behaviors relevant to the p r e v e n t i o n of medical conditions, s u c h as ischem i c heart disease, and was population-based. In addition, attitudes and changes in health b e h a v i o r over the p r e v i o u s two years w e r e assessed. The results are presented here.
METHODS Data w e r e c o l l e c t e d through in-person interviews regarding changes in health behaviors in the United States, the United Kingdom, and France. Interviews w e r e c o n d u c t e d b e t w e e n June and N o v e m b e r 1988. Before discussing the content of the interviews, the strategy for drawing samples for the evaluation is examined b e l o w .
Sample Selection The s a m p l e was restricted to persons aged 16 to 50 years. Nationally representative samples w e r e drawn r a n d o m l y in the United States, the United Kingdom, and France. Persons living in Corsica, Northern Ireland, Alaska, Hawaii, and n o n c o n t i g u o u s territories w e r e excluded. Also, prisoners, hospitalized patients, and persons residing in religious or educational institutions w e r e e x c l u d e d . Sampling and fielding of the in-person interviews w e r e p e r f o r m e d by Louis Harris and Associates. Since one of the p u r p o s e s of c o n d u c t i n g the surveys was to d e t e r m i n e k n o w l e d g e and practices regarding risk factors for AIDS, certain d e m o g r a p h i c characteristics w e r e o v e r s a m p l e d (e.g., men, the young, ethnic minorities, and the u n d e r e d u c a t e d ) . The sampling a p p r o a c h used in each c o u n t r y i n c l u d e d stratified clusters o f persons in urban and n o n u r b a n locations, and g e o g r a p h i c areas w e r e identified to represent the d e m o g r a p h i c characteristics for oversampling. For example, in the United States census tracts with high p r o p o r t i o n s of minorities w e r e o v e r s a m p l e d . In each country, the o v e r s a m p l i n g strategy ensured greater reliability of the data, thus, small n u m b e r s w e r e avoided for specific characteristics (e.g., race). Clusters of households w e r e r a n d o m l y s a m p l e d within g e o g r a p h i c areas for c o n d u c t i n g the survey. Selection of households for c o n d u c t i n g interviews was p e r f o r m e d by obtaining an u p d a t e d list of the n u m b e r o f o c c u p i e d dwelling units (ODUs) within g e o g r a p h i c areas. The n u m b e r of ODUs was divided b y the n u m b e r of interviews to be c o n d u c t e d in a particular geographic area, and an interval b e t w e e n ODUs was comp u t e d for sampling. Taken from a r a n d o m l y selected c o r n e r of a block, or b l o c k equivalent, this distance r e p r e s e n t e d the first ODU at w h i c h an interview was c o n d u c t e d . The remaining ODUs in the cluster w e r e
d e t e r m i n e d b y adding the c o m p u t e d interval b e t w e e n units until ten ODUs w e r e identified. Housing units with at least one resident b e t w e e n the ages of 16 and 50 w e r e eligible. In housing units w h e r e m o r e than one resident was eligible, the p e r s o n with the most recent birthday was interviewed. Interviewers m a d e at least four attempts to obtain interviews; other households within g e o g r a p h i c clusters w e r e selected to replace those r e s p o n d e n t s w h o refused or c o u l d not be contacted. R e p l a c e m e n t households w e r e also selected randomly using the m e t h o d described above.
Survey The survey instrument i n c l u d e d d e m o g r a p h i c information a b o u t age, gender, marital status, race, and education. Because of some important differences a m o n g the three countries for some of these characteristics, the interviews w e r e a p p r o p r i a t e l y modified. For example, minority status and educational levels differ from c o u n t r y to country. Nonetheless, differences a m o n g the countries in minority status (described in the Results section) and educational levels w e r e determined. For educational level, three categories w e r e used: less than a secondary education, c o m p l e t e d a secondary education, and c o m p l e t e d a university education. Questions regarding current health behaviors inc l u d e d information about current weight and height as well as diet, alcohol c o n s u m p t i o n , exercise, and cigarette use. The f r e q u e n c y of strenuous exercise was det e r m i n e d b y asking h o w often respondents participated in heavy exertion for at least 20 minutes a day. Quetelet indices ( k g / c m 2) w e r e c o n s t r u c t e d to e x a m i n e differences in b o d y mass. Attitudes and beliefs regarding health behaviors and their c o n s e q u e n c e s w e r e also addressed. Attitudes included fear and responsibility for d e v e l o p i n g medical illness, c o n c e r n for health, attitudes a b o u t lifestyle, and beliefs regarding the role of chance in health and illness. Beliefs w e r e also solicited about the role of selected risk factors for d e v e l o p i n g cardiovascular disease (e.g., smoking, exercise). Finally, survey items included health b e h a v i o r changes over the previous two years for smoking, exercising, alcohol c o n s u m p t i o n , diet, and weight. All questions regarding health behaviors w e r e p o s e d b y the interv i e w e r in person."
Analysis The data w e r e analyzed using the Statistical Analysis System. 9 The three countries w e r e c o m p a r e d for personal characteristics, health behaviors, attitudes toward health, and changes in health behaviors over the
•The instrument used for determining health behaviors and attitudes is available u p o n request.
JOURNALOFGENERALINTERNALMEDICINE,Volume 7 (November/December), 1992
previous two years. Mean values and standard errors are reported for continuous variables and percentiles. Values were i m p u t e d for some variables with missing data by randomly assigning values in p r o p o r t i o n to their observed distribution or, in the case of continuous variables, by assigning the mean; missing data w e r e present for less than 1.5% of cases. Weights w e r e assigned to reflect true proportions in the populations to provide appropriate national estimates. Since oversamples of men and minorities were drawn, weights were used to adjust for key variables (education, race, age, and gender) to reflect their actual proportions in the populations of the three countries. Comparisons b e t w e e n the three countries regarding personal characteristics were p e r f o r m e d using analysis of variance. Differences b e t w e e n the United States, the United Kingdom, and France for health behaviors, attitudes, and changes in health behaviors w e r e determined by comparing proportions using a general linear model. For these analyses, since multiple comparisons were performed, Bonferroni's adjustment was u s e d ) ° Statistical significance was established at p < 0.002, to correct for multiple comparisons, o~ = O.05/n, w h e r e the n u m b e r of comparisons equaled 25. The n u m b e r o f comparisons included each nondemographic variable examined; for this purpose, each analysis across countries (e.g., United States vs. United Kingdom, United States vs. France, United Kingdom vs. France) was considered a single comparison. Confidence intervals for the true differences in proportions b e t w e e n countries were estimated by P2 -- Pl + Z4[p~(1 -- p l ) / n t + p2(1 -- p2)/n2] w h e r e P2-----the p r o p o r t i o n of respondents in the United States with a particular health behavior and Pl = the proportion of respondents in the United Kingdom or France w h o r e p o r t e d the same behavior. 11 Again, to control for the chance of finding a significant
617
difference through multiple comparisons, confidence intervals w e r e adjusted for 25 comparisons. Finally, a multivariate linear model was also used to control for differences b e t w e e n countries regarding changes in health behaviors over the previous two years. The dep e n d e n t variable included changes in u p to six different health behaviors (decrease in smoking, increase in f r e q u e n c y of exercise, decrease in alcohol consumption, decrease in egg consumption, decrease in red meat consumption, and decrease in weight). Confidence intervals (95%) for pairwise differences between means for i n d e p e n d e n t variables were also determined. Two scales were used to analyze differences between countries in attitudes toward fear and responsibility of developing disease. One attitude scale considered the respondent's fear of developing specific medical illnesses (diabetes, cancer, AIDS, crippling disease, venereal disease, mental illness, and heart attack), while the other addressed personal responsibility for acquiring selected diseases and conditions (diabetes, cancer, crippling disease, mental illness, and heart attack). The factor loadings for individual items in the scale regarding fear of disease w e r e high (-> 0.7) for most elements (Cronbach's o~ = 0 . 8 8 ) . The factor loadings for the scale dealing with personal responsibility w e r e slightly lower (-> 0,5), but still within acceptable standards (Cronbach's o~ = 0.66).
RESULTS Interviews were c o m p l e t e d for 1,940 persons in the United States, 1,833 persons in the United Kingdom, and 2,294 persons in France. Overall response rates were: 67% in the United States, 65% in the United Kingdom, and 71% in France. Response rates for oversampled segments (e.g., minorities) were slightly less than for the sample as a whole. Nonrespondents inc l u d e d those persons w h o refused to be interviewed as
TABLE 1 • DemographicCharacteristicsof the SurveyedPopulationsin the United States, the United Kingdom, and France* United States
United Kingdom
(n = 1,940)
(n = 1,833)
Age--mean
32.0 years (0.2 years)
31.8 years (0.2 years)
31.5 years (0.2 years)
Male
49.8% (1.1%)
50.4% (1.2%)
50.1% (1.0%)
Minorityt
24.3% (1.0%)
6.8% (0.6%)
12.3% (0.7%)
->Collegeeducationt
18.6% (0.9%)
16.2% (0.9%)
24.6% (0.9%)
Marriedt
54.1% (1.1%)
59.6% (1.1%)
62.5% ( 1.0%)
Very good or excellenthealtht
69.4% (1.1%)
59.8% (1.1%)
81.3% (0.8%)
Employedt
80.9% (0.9%)
70.5% (1.1%)
63.6% ( 1.0%)
*Standard errors of the mean or of the percentageare provided in parentheses. t Significantlydifferent across all three countries, p < 0.05, by analysisof variance.
France (n = 2,294)
Retchin eta/., HEALTHBEHAVIORCHANGES
618
E 0
0"I
(:5
o 0
~:
,~ O~
~
(~0
~
o d
I
u~
I
~o
I
o
8
t",l
I
I
d ~
v
~o 06
~. ~
'~ d
~d
I
I
~
I
~4 • I/]
o
0 i/~
t/1
0
U'I
g:
~
d
,~
~"
~
o
,,6
t
1992;7:615 -622. NUMEROUS EPIDEMIOLOGICSTUDIESover the last several decades have demonstrated the relationship of selected health behaviors to physical health status and disease.t. 2 Since that time, major health p r o m o t i o n campaigns have been introduced, and modification of Received from the Project HOPE Center for Health Affairs, ChevyChase, Maryland (SMR,JAW),the Department of Internal Medicine, Medical College of Virginia, Virginia CommonwealthUniversity, Richmond, Virginia (SMR), the Institut de la Sant~ et de la Recherche M~dicale, Paris, France (A-JV), and the School of Public Health, Universityof Illinois-Chicago, Chicago, Illinois (GLA). Presented in part at the annual meeting of the Societyof General Internal Medicine, May 1- 3, 1991, Seattle, Washington. Address correspondence and reprint requests to Dr. Retchin: Box 287, MCVStation, Richmond, VA23298.
health behaviors has b e c o m e a major health care focus in the United States. 3 Distinguished authorities, such as the Institute of Medicine, have endorsed the role of health behaviors in the cause o f specific medical illnesses, particularly ischemic heart disease. 4 Finally, considerable media attention has been concentrated on health and illness, and as a result of these and other efforts, there have been substantial health behavior changes, s Although there is evidence that health behaviors have c h a n g e d substantially in recent years, less is k n o w n about international trends and comparisons a m o n g countries. International disparities in health behaviors are important to examine for several reasons. First, differences in health behaviors b e t w e e n countries are important because they may provide insights regarding changing patterns of disease. For example, although meaningful declines in ischemic heart disease have been observed over the last decade and a half in Northern and Western Europe ( 1 0 - 14%), by far the largest reductions have taken place in the United States -> 25%). 6 If health behaviors are at least partially responsible for observed variations a m o n g countries in rates of decline, then differences in corresponding health behaviors w o u l d be e x p e c t e d to parallel these variations. Second, changes in health behaviors in different societies may be the result of distinguishing sociocultural factors, 7, 8 and these factors are often difficult to investigate within countries. Since there are notable cultural influences that affect these factors, studies that involve international comparisons are important. As an example, following p r o f o u n d secular changes in the United States during the '50s and '60s, the p r o p o r t i o n o f w o m e n w h o s m o k e d rose considerably. As a result, rates of lung cancer in w o m e n are rising in the United States disproportionate to rates in some other countries. Last, the dissemination of scientific evidence regarding health p r o m o t i o n is strongly affected by the medical c o m m u n i t y , p u b l i c receptivity, and other factors that are peculiar to each society. Media interest and exposure for p u b l i c health issues in different countries are d e t e r m i n e d by p o p u l a r concerns, cultural environments, and g o v e r n m e n t regulations. Because these cultural characteristics are difficult to examine within countries, international differences in health attitudes may provide important opportunities for examining o u t c o m e s as a result of the different exposures to these influences. As part of an international study of behavior 615
616
Retchin et aL, HEALTHBEHAVIORCHANGES
changes related to the transmission of AIDS and cardiovascular disease, w e c o n d u c t e d a survey in the United States, the United Kingdom, and France. This survey i n c l u d e d questions regarding health behaviors relevant to the p r e v e n t i o n of medical conditions, s u c h as ischem i c heart disease, and was population-based. In addition, attitudes and changes in health b e h a v i o r over the p r e v i o u s two years w e r e assessed. The results are presented here.
METHODS Data w e r e c o l l e c t e d through in-person interviews regarding changes in health behaviors in the United States, the United Kingdom, and France. Interviews w e r e c o n d u c t e d b e t w e e n June and N o v e m b e r 1988. Before discussing the content of the interviews, the strategy for drawing samples for the evaluation is examined b e l o w .
Sample Selection The s a m p l e was restricted to persons aged 16 to 50 years. Nationally representative samples w e r e drawn r a n d o m l y in the United States, the United Kingdom, and France. Persons living in Corsica, Northern Ireland, Alaska, Hawaii, and n o n c o n t i g u o u s territories w e r e excluded. Also, prisoners, hospitalized patients, and persons residing in religious or educational institutions w e r e e x c l u d e d . Sampling and fielding of the in-person interviews w e r e p e r f o r m e d by Louis Harris and Associates. Since one of the p u r p o s e s of c o n d u c t i n g the surveys was to d e t e r m i n e k n o w l e d g e and practices regarding risk factors for AIDS, certain d e m o g r a p h i c characteristics w e r e o v e r s a m p l e d (e.g., men, the young, ethnic minorities, and the u n d e r e d u c a t e d ) . The sampling a p p r o a c h used in each c o u n t r y i n c l u d e d stratified clusters o f persons in urban and n o n u r b a n locations, and g e o g r a p h i c areas w e r e identified to represent the d e m o g r a p h i c characteristics for oversampling. For example, in the United States census tracts with high p r o p o r t i o n s of minorities w e r e o v e r s a m p l e d . In each country, the o v e r s a m p l i n g strategy ensured greater reliability of the data, thus, small n u m b e r s w e r e avoided for specific characteristics (e.g., race). Clusters of households w e r e r a n d o m l y s a m p l e d within g e o g r a p h i c areas for c o n d u c t i n g the survey. Selection of households for c o n d u c t i n g interviews was p e r f o r m e d by obtaining an u p d a t e d list of the n u m b e r o f o c c u p i e d dwelling units (ODUs) within g e o g r a p h i c areas. The n u m b e r of ODUs was divided b y the n u m b e r of interviews to be c o n d u c t e d in a particular geographic area, and an interval b e t w e e n ODUs was comp u t e d for sampling. Taken from a r a n d o m l y selected c o r n e r of a block, or b l o c k equivalent, this distance r e p r e s e n t e d the first ODU at w h i c h an interview was c o n d u c t e d . The remaining ODUs in the cluster w e r e
d e t e r m i n e d b y adding the c o m p u t e d interval b e t w e e n units until ten ODUs w e r e identified. Housing units with at least one resident b e t w e e n the ages of 16 and 50 w e r e eligible. In housing units w h e r e m o r e than one resident was eligible, the p e r s o n with the most recent birthday was interviewed. Interviewers m a d e at least four attempts to obtain interviews; other households within g e o g r a p h i c clusters w e r e selected to replace those r e s p o n d e n t s w h o refused or c o u l d not be contacted. R e p l a c e m e n t households w e r e also selected randomly using the m e t h o d described above.
Survey The survey instrument i n c l u d e d d e m o g r a p h i c information a b o u t age, gender, marital status, race, and education. Because of some important differences a m o n g the three countries for some of these characteristics, the interviews w e r e a p p r o p r i a t e l y modified. For example, minority status and educational levels differ from c o u n t r y to country. Nonetheless, differences a m o n g the countries in minority status (described in the Results section) and educational levels w e r e determined. For educational level, three categories w e r e used: less than a secondary education, c o m p l e t e d a secondary education, and c o m p l e t e d a university education. Questions regarding current health behaviors inc l u d e d information about current weight and height as well as diet, alcohol c o n s u m p t i o n , exercise, and cigarette use. The f r e q u e n c y of strenuous exercise was det e r m i n e d b y asking h o w often respondents participated in heavy exertion for at least 20 minutes a day. Quetelet indices ( k g / c m 2) w e r e c o n s t r u c t e d to e x a m i n e differences in b o d y mass. Attitudes and beliefs regarding health behaviors and their c o n s e q u e n c e s w e r e also addressed. Attitudes included fear and responsibility for d e v e l o p i n g medical illness, c o n c e r n for health, attitudes a b o u t lifestyle, and beliefs regarding the role of chance in health and illness. Beliefs w e r e also solicited about the role of selected risk factors for d e v e l o p i n g cardiovascular disease (e.g., smoking, exercise). Finally, survey items included health b e h a v i o r changes over the previous two years for smoking, exercising, alcohol c o n s u m p t i o n , diet, and weight. All questions regarding health behaviors w e r e p o s e d b y the interv i e w e r in person."
Analysis The data w e r e analyzed using the Statistical Analysis System. 9 The three countries w e r e c o m p a r e d for personal characteristics, health behaviors, attitudes toward health, and changes in health behaviors over the
•The instrument used for determining health behaviors and attitudes is available u p o n request.
JOURNALOFGENERALINTERNALMEDICINE,Volume 7 (November/December), 1992
previous two years. Mean values and standard errors are reported for continuous variables and percentiles. Values were i m p u t e d for some variables with missing data by randomly assigning values in p r o p o r t i o n to their observed distribution or, in the case of continuous variables, by assigning the mean; missing data w e r e present for less than 1.5% of cases. Weights w e r e assigned to reflect true proportions in the populations to provide appropriate national estimates. Since oversamples of men and minorities were drawn, weights were used to adjust for key variables (education, race, age, and gender) to reflect their actual proportions in the populations of the three countries. Comparisons b e t w e e n the three countries regarding personal characteristics were p e r f o r m e d using analysis of variance. Differences b e t w e e n the United States, the United Kingdom, and France for health behaviors, attitudes, and changes in health behaviors w e r e determined by comparing proportions using a general linear model. For these analyses, since multiple comparisons were performed, Bonferroni's adjustment was u s e d ) ° Statistical significance was established at p < 0.002, to correct for multiple comparisons, o~ = O.05/n, w h e r e the n u m b e r of comparisons equaled 25. The n u m b e r o f comparisons included each nondemographic variable examined; for this purpose, each analysis across countries (e.g., United States vs. United Kingdom, United States vs. France, United Kingdom vs. France) was considered a single comparison. Confidence intervals for the true differences in proportions b e t w e e n countries were estimated by P2 -- Pl + Z4[p~(1 -- p l ) / n t + p2(1 -- p2)/n2] w h e r e P2-----the p r o p o r t i o n of respondents in the United States with a particular health behavior and Pl = the proportion of respondents in the United Kingdom or France w h o r e p o r t e d the same behavior. 11 Again, to control for the chance of finding a significant
617
difference through multiple comparisons, confidence intervals w e r e adjusted for 25 comparisons. Finally, a multivariate linear model was also used to control for differences b e t w e e n countries regarding changes in health behaviors over the previous two years. The dep e n d e n t variable included changes in u p to six different health behaviors (decrease in smoking, increase in f r e q u e n c y of exercise, decrease in alcohol consumption, decrease in egg consumption, decrease in red meat consumption, and decrease in weight). Confidence intervals (95%) for pairwise differences between means for i n d e p e n d e n t variables were also determined. Two scales were used to analyze differences between countries in attitudes toward fear and responsibility of developing disease. One attitude scale considered the respondent's fear of developing specific medical illnesses (diabetes, cancer, AIDS, crippling disease, venereal disease, mental illness, and heart attack), while the other addressed personal responsibility for acquiring selected diseases and conditions (diabetes, cancer, crippling disease, mental illness, and heart attack). The factor loadings for individual items in the scale regarding fear of disease w e r e high (-> 0.7) for most elements (Cronbach's o~ = 0 . 8 8 ) . The factor loadings for the scale dealing with personal responsibility w e r e slightly lower (-> 0,5), but still within acceptable standards (Cronbach's o~ = 0.66).
RESULTS Interviews were c o m p l e t e d for 1,940 persons in the United States, 1,833 persons in the United Kingdom, and 2,294 persons in France. Overall response rates were: 67% in the United States, 65% in the United Kingdom, and 71% in France. Response rates for oversampled segments (e.g., minorities) were slightly less than for the sample as a whole. Nonrespondents inc l u d e d those persons w h o refused to be interviewed as
TABLE 1 • DemographicCharacteristicsof the SurveyedPopulationsin the United States, the United Kingdom, and France* United States
United Kingdom
(n = 1,940)
(n = 1,833)
Age--mean
32.0 years (0.2 years)
31.8 years (0.2 years)
31.5 years (0.2 years)
Male
49.8% (1.1%)
50.4% (1.2%)
50.1% (1.0%)
Minorityt
24.3% (1.0%)
6.8% (0.6%)
12.3% (0.7%)
->Collegeeducationt
18.6% (0.9%)
16.2% (0.9%)
24.6% (0.9%)
Marriedt
54.1% (1.1%)
59.6% (1.1%)
62.5% ( 1.0%)
Very good or excellenthealtht
69.4% (1.1%)
59.8% (1.1%)
81.3% (0.8%)
Employedt
80.9% (0.9%)
70.5% (1.1%)
63.6% ( 1.0%)
*Standard errors of the mean or of the percentageare provided in parentheses. t Significantlydifferent across all three countries, p < 0.05, by analysisof variance.
France (n = 2,294)
Retchin eta/., HEALTHBEHAVIORCHANGES
618
E 0
0"I
(:5
o 0
~:
,~ O~
~
(~0
~
o d
I
u~
I
~o
I
o
8
t",l
I
I
d ~
v
~o 06
~. ~
'~ d
~d
I
I
~
I
~4 • I/]
o
0 i/~
t/1
0
U'I
g:
~
d
,~
~"
~
o
,,6
t
