Smoking among Mexican Americans: A Three-Generation Study - NCBI

1 downloads 0 Views 746KB Size Report
women. However, the percentage oflight smokers was considerably. Introduction. Low rates of smoking have been thought to be a major factor behind the lower ...
Smoking among Mexican Americans: A Three-Generation Study KYRIAKOS S. MARKIDES, PHD, JEANNINE COREIL, PHD, AND LAURA A. RAY, MPA Abstract: Smoking behavior was investigated with data from a three-generation study of Mexican Americans in San Antonio, Texas. Rates of smoking among the men were found to be higher than rates for other White men obtained in previous studies, while rates for women were slightly lower than those reported for other White women. However, the percentage of light smokers was considerably

Introduction Low rates of smoking have been thought to be a major factor behind the lower lung cancer (and other smokingrelated cancers) rate among Mexican Americans, particularly the men."13 While a number of studies show that Mexican Americans have lower rates of smoking than other Whites and Blacks,A' few studies have looked at the predictors of smoking behavior in this population. For example, the effects of age, sex, generation, level of acculturation, place of origin, and other sociocultural factors have received little attention. Marcus and Crane7 recently reviewed the literature on smoking among US Hispanics and pointed out that various studies reveal consistent sex differences in smoking prevalence relative to non-Hispanic Whites. While Hispanic men have higher rates of smoking than other Whites and Blacks, the women have lower rates than the other groups. The above pattern was observed in the 1979 and 1980 National Health Interview Surveys,7 a California-wide survey,8 a Texas-wide survey,9 a study of three socioeconomically distinct San Antonio neighborhoods,* and in data collected by the New Mexico Lung Cancer Case Control Study.'0 Further analysis of the National Health Interview Survey data by type of Spanish Origin showed that the gap by sex was more pronounced in the Mexican origin population. In view of the disparity between current smoking patterns and rates of cancer and other smoking-related diseases among Hispanics, it has been suggested that smoking has increased from previously lower rates, along with greater acculturation. " Some study findings, however, do not fit the pattern just described. Samet, et al, 12 for example, found similar smoking rates among Hispanic and other White males and females in a New Mexico county. Also, data from the 1981-83 Behavioral Risk Factor Surveys conducted in 28 states and the District of Columbia showed that slightly fewer Hispanic men than Black and other White men were current smokers, but the differences were small. Hispanic women, on the other hand, were substantially less likely to be smokers than other women. 13 While a few studies have given attention to age differences in Hispanic smoking patterns,'0"12 none have systematically investigated the effects of generation and familial influence. This paper presents data on smoking collected recently in a three-generation study of Mexican Americans in San Antonio. Emphasis is given to sex, age/generational, and From the Department of Preventive Medicine and Community Health,

University of Texas Medical Branch, Galveston, TX 77550. Address reprint requests to Dr. Markides, Associate Professor, at that address. This paper, submitted to the Journal October 30, 1985, was revised and accepted for

publication November 24, 1986.

© 1987 American Journal of Public Health 0090-0036/87$1.50

708

higher among Mexican Americans than among other groups. There was no evidence that acculturation was a consistent predictor of greater likelihood of smoking. However, there was evidence that the

smoking behavior of younger Mexican Americans, particularly the women, was associated with the smoking behavior of their parents. (Am J Public Health 1987; 77:708-711.) acculturation differences in smoking behavior. In addition, since the three generations are intergenerationally linked, we investigate evidence of familial aggregation of smoking behavior by conducting within family analysis. Methods

The data were collected in 1981 and 1982 through home interviews with 1,125 Mexican Americans residing in the San Antonio, Texas area. Older Mexican Americans (aged 65-80) with three-generationally linked families located in the San Antonio area were selected first, using area probability sampling.'4 Information on their children and ever-married adult grandchildren (18 years +) was subsequently obtained. Three-generation lineages including the older person, a middle-aged child, and a grandchild (all in the same family line) were selected randomly. Approximately two-thirds of the sample were women, owing largely to their greater longevity in the older generation, and their tendency to marry and have children at younger ages than men. The mean ages were 74, 49, and 26 in the three-generations (see footnote to Table 1), with minimal overlap in age among successive

generations. Respondents were first asked whether they currently smoked and whether they smoked cigarettes, a pipe, or cigars. Smokers were asked how many cigarettes, pipefulls or cigars they smoked per day, and whether they ever tried to quit smoking. (Only nine respondents reported smoking a pipe or cigars.) Nonsmokers were asked whether they had smoked in the past and, if so, when they had stopped. These data were categorized into two dichotomous dependent variables: current smoking (current smokers = 1; others = 0), and ever having smoked (ever smoked = 1; never = 0). The independent variables examined in relation to the smoking measures included sex, age, generation, marital

status, employment status, sex-role orientation, and three indices of acculturation.'5 All analyses were performed separately by sex. Because of the low prevalence of smoking in the elderly group, the analyses were confined to the middle and younger generations. Age, education, and marital status were included as control variables for both sexes. For females, we also looked at the influence of employment status (employed = 1) and traditional sex-role orientation (measured by a seven-item scale).'6 The acculturation indices included: 1) a five-point scale measuring ethnicity of friends; 2) a four-item language scale measuring extent of Spanish used in TV programs watched, in radio programs listened to, and in conversation with spouse and friends; and 3) a traditional values scale measuring the importance attributed to Mexican heritage (e.g., celebrating certain holidays, having Mexican American friends). The latter two scales had alpha reliability coeffiAJPH June 1987, Vol. 77, No. 6

SMOKING AMONG MEXICAN AMERICANS TABLE 1-Current Smokers, Ex-Smokers and Persons Who Never Smoked by Generation and Sex

TABLE 2-Results of Multiple Logistic Regression Analysis of Smoking Status Variables (Middle and Younger Generations Only)

Smoking Status Current

Ex

Current Smoker Never

Male

Total Variables

Generation/Sex

No.

(%)

No.

Male Younger Generation 68 (47.9) 14 Middle Generation 64 (50.0) 26 Older Generation 39 (37.9) 42 171 (45.8) 82 TOTAL Female Younger Generation 55 (24.2) 27 Middle Generation 68 (28.0) 29 Older Generation 45 (17.1) 54 TOTAL 168 (22.9) 110

(%)

No.

(%)

No.

(%)

(9.9) 60 (20.3) 38 (40.8) 22 (22.0) 120

(42.2) (29.7) (21.3) (32.2)

142 128 103 373

(100) (100) (100) (100)

(11.8) (11.9) (20.5) (15.0)

(64.0) (60.1) (62.4) (62.1)

228 243 263 734

(100) (100) (100) (100)

146 146 164 456

NOTE: Mean age for younger generation was 26 years (SD = 4.8, range = 18-42); for middle generation was 49 years (SD = 6. 1, range = 32-68); and for older generation was 74 (SD = 4.7, range = 65-80).

cients between 0.75 and 0.84 for the different generations.'5 These three indices are considered to be conceptually distinct dimensions of acculturation.'7 To test the acculturation hypothesis, multiple logistic regression analysis was performed with the smoking status variables. We also conducted within-family analysis to test for evidence of intergenerational transmission of smoking behavior. To also examine the differential influence of fathers' versus mothers' smoking, we conducted the analysis by type of sex-linkage. Since preliminary analysis showed few significant within-family associations between the older and middle generations, the reported results are confined to the influence of middle-aged parents' smoking behavior on the smoking of their children. Multiple logistic regression analysis was used to determine intergenerational effects.'8 Results Table 1 shows the distribution of smokers, ex-smokers, and persons who never smoked by generation and sex. The per cent of current smokers is lowest in the older generation, and females currently smoked at about half the rates of men in all generations. Even though these rates are relatively high, particularly among men, the sample contains a relatively large number of light smokers (10 or fewer cigarettes per day): among males 79.7 per cent of the smokers in the younger generation, 61.7 per cent in the middle, and 68.4 per cent in the older generation. The corresponding percentages were even higher among women. The results of our regression analysis with the two younger generations combined did not show consistent evidence that acculturation is related to greater likelihood of smoking among either men or womien (Table 2). When predicting current smoking among men, age was negatively related to smoking as was the language scale: men who used English more frequently were less likely to be smokers, a finding which is not consistent with the acculturation hypothesis. However, ethnicity of friends was positively related to smoking in that men who have more Anglo friends were more likely to have ever smoked. Women who were married were less likely to have ever smoked. Women who scored higher on the traditional values scale were more likely to smoke or to have ever smoked, which does not support the acculturation hypothesis.

AJPH June 1987, Vol. 77, No. 6

Age

(N

=

259)

Ever Smoked

Female

(453)

Male (N

=

259)

Female (N = 444)

-.028

-.010

.002

Marital Status

(.013)

(.010)

(.012)

-.278

-.228

(.009)

-.094

(.188)

-.233

(.120)

(.194)

Education

(.112)

-.035

-.033

-.034

-.032

Language Scale

(.048)

(.042)

(.050)

-.116

.007

(.039)

-.089

-.009

(.053)

(.040)

(.055)

.030 (.021) .344 (.185) -.233 (.188) .051

.042 (.018) -.066 (.160) -.127

(.037)

.005

.011 (.022) .482 (.201) -.070 (.199) .050

-.029

(.043)

(.036)

(.045)

(.033)

Traditional Values Scale

Ethnicity of Friends Employment Sex Role Orientation

(.117)

-.010

.033

(.016) .045 (.144) .026

(.106)

NOTE: Logistic coefficients are shown with standard errors in parentheses.

To explore further the acculturation hypothesis, we decided to regress the ethnicity of co-workers ("How many of your co-workers are Anglos?" None to all, scored 1-4) on the smoking variables for people who were currently employed. We found that women who had more Anglo coworkers were more likely to have ever smoked than women who had more Mexican American co-workers (other things equal). Table 3 shows results of a series of multiple logistic regressions predicting the smoking behavior of the younger generation from the smoking behavior of their parents. Odds ratios presented were adjusted for the younger generation's level of education, sex role orientation, and ethnicity of friends. We chose only three controls because of small Ns when the sample is broken down by type of sex-linkage. The equations predicting the smoking status of all younger men and all younger women were also adjusted by gender of their parents. The equations including all cases together (men and women) also include an adjustment for the gender of the younger generation. In all these analyses, female was coded as 1 and male as 0. Table 3 shows that even after controls, middle-aged parents' smoking behavior is a significant predictor of the smoking behavior of the young adult children. For example, younger people whose parents are smokers were 1.66 times more likely to be smokers than younger people whose parents are not smokers. Most odds ratios are clearly greater than 1.0. Finally, the father-daughter and mother-daughter associations seem slightly stronger than the father-son and mother-son associations. Discussion The prevalence estimates in our study should be interpreted with caution because of the unique nature of our sample which does not represent the Mexican American community at large but persons from three-generations families as we have defined them. Yet our findings are consistent with Marcus and Crane's compilation of recent evidence.7 Our rates for males are higher than those found in most studies of Spanish Origin groups,7"3 but similar to those found for Hispanic men in New Mexico.'0 709

MARKIDES, ET AL. TABLE 3-Odds Ratios Derived from Multiple Logistic Regression Analysis of Influence of Parents' (Middle Generation) Smoking on Children's (Younger Generation) Smoking by Type of Sex Linkage Younger Generation

Current Smoker

Ever Smoked

Odds Ratio (95% Confidence

Middle Generation

Odds Ratio (95% Confidence

N

Interval)

N

367

362

363

1.66 (1.30, 2.14) 1.54 (1.12, 1.97)

Current Smoker

142

1.66 (1.13, 2.43)

142

Ever Smoked

141

1.24 (.85, 1.80)

141

1.73 (1.23, 2.44) 1.87 (1.32, 2.65)

220 217

1.60 (.87, 2.93)

58

1.10

58

.98 (.50, 1.93)

66

1.48 (1.22, 2.67) 2.06 (0.99, 4.29)

All Cases Current Smoker

Ever Smoked

Interval) 1.42

(0.79, 1.80) 363

1.43 (1.14,1.81)

All Males (younger)

All Females (younger) Current Smoker

Ever Smoked

Father-Son Current Smoker Ever Smoked

225 222

58 58 67

2.21

Mother-Daughter Current Smoker Ever Smoked Mother-Son Current Smoker

66

158

1.67 (1.12, 2.50)

154

1.72

151

83

1.28 (.82, 2.01)

1.53 1.67

(1.17, 2.38) 84

(1.01, 2.77)

Ever Smoked

1.42

(1.04, 2.24)

(1.18, 2.52) 1.68

1.65

(.77, 2.61)

67

84

1.45

(1.22, 2.24)

-

155

1.57

(1.09, 2.26) 1.31 83

(1.23, 2.11)

NOTE: Odds ratios are adjusted for the third generation's level of education, sex role orientation, and ethnicity of friends. The equations containing all cases are also adjusted for gender in both generations. Equations including only males or females in the third generation include an adjustment for the gender of the parent (in all cases, female = 1; male = 0)

As in previous studies, few of the Mexican American women in our study were smokers. Although the figures by age were slightly lower than found in the general population,'3 they are slightly higher than those found for Hispanic women in previous studies.7-'2 The proporiton of light smokers in our sample was higher than that found in the general population and is consistent with other studies of Hispanics.7"2 That the Mexican Americans in our study were clearly more likely to be smokers than Mexican Americans and other Hispanics in other studies led us to speculate about the unique nature of our three-generation sample. We hypothesized a possible acculturation effect, particularly since virtually all of our middle and younger generation subjects were born in the United States. However, the analyses failed to show any consistent influence of several acculturation indicators on smoking rates. Our findings do not lend support to the hypothesis that increased smoking rates reflect a part of 710

smoking behavior.'9'20 Even though we have not been able to specify well the influence of a number of variables on the smoking behavior of Mexican Americans, our findings clearly indicate that smoking rates among Mexican Americans in San Antonio are high, particularly among men. It is also likely that these high smoking rates will translate into rising rates of cancer and other smoking-related diseases in the near future.7"

(1.06, 1.98)

(1.03, 4.63) Ever Smoked

1.21

(.83, 1.77)

(.56, 2.14) Father-Daughter Current Smoker

1.49

(1.01, 2.20)

a general process of acculturation to the lifestyle of the larger society. "I Although acculturation measures failed to consistently influence smoking rates, within family analysis indicated that familial transmission might be important between the middle generation and their children. The fact that the associations suggest that younger women are more likely to be influenced by the smoking behavior of their parents than are younger men, suggests a greater role of the family in socializing women; the influence of extrafamilial factors, such as peer group pressures, is likely to be stronger among men. This notion finds some support from the literature on adolescent

ACKNOWLEDGMENTS

This study was supported by grant ROl AG04170 from the National Institute on Aging and by grants from the Hogg Foundation for Mental Health and the American Heart Association (Texas Affiliate). We would like to thank Richard DeFrank, Sally Vernon, and Marjorie Speers for useful comments on earlier drafts.

REFERENCES

I. Lee ES, Roberts RE, Labarthe DR: Excess and deficit lung cancer mortality in three ethnic groups in Texas. Cancer 1976; 38:2551-2556. 2. Menck H, Henderson B, Pike M, et al: Cancer incidence in the MexicanAmerican. JNCI 1975; 55:531-536. 3. Samet JM, Key CR, Kutvirt DM, Wiggins CL: Respiratory disease mortality in New Mexico's American Indians and Hispanics. Am J Public Health 1980; 70:492-497. 4. Roberts RE, Lee ES: Health practices among Mexican-Americans: Further evidence from the Human Population Laboratory studies. Prev Med 1980; 9:675-688. 5. Holck S, Warren C, Rochat C, Smith J: Lung cancer mortality and smoking habits: Mexican-American women. Am J Public Health 1982; 72:38-42. 6. National Center for Health Statistics: Health practices among adults: United States, 1977. (Advance Data No. 64.) HHS Pub No. (PHS) 81-1250. Hyattsville, MD: NCHS 1980. 7. Marcus AC, Crane LA: Smoking behavior among US Latinos: an emerging challenge for public health. (commentary) Am J Public Health 1985; 75:169-172. 8. Igra A, Stavig GR, Leonard AR: Hypertension and related health problems in California: Results for the 1979 California Hypertension Survey. Sacramento: California Department of Health Services, Hypertension Control Program, n.d. 9. Texas Department of Health: Texas Behavioral Risk Factor Survey: Ethnic Differences. Report by the TDH, Bureau of State Health Planning and Resource Development, Austin, 1983. 10. Humble CG, Samet JM, Pathak RR, Skipper BJ: Cigarette smoking and lung cancer in New Mexico's Hispanics and Anglos. Am J Public Health 1985; 75:145-148. 11. Savitz, DA: Changes in Spanish surname cancer rates relative to other Whites in the Denver area 1969-71 to 1979-81. Am J Public Health 1986;

76:1210-1215.

12. Samet JM, Schrag SD, Howard CA, et al: Respiratory disease in a New Mexico population sample of Hispanic and Non-Hispanic Whites. Am Rev Respir Dis 1982; 125:152-157. 13. Remington PL, Forman MR, Gentry M, et al: Current smoking trends in the United States: The 1981-83 Behavioral Risk Factor Surveys. JAMA 1985; 253:2975-2978. 14. Markides KS, Hoppe SK, Martin HW, Timbers DM: Sample representativeness in a three-generation study of Mexican Americans. J Marr Fam

1983; 45:911-916.

15. Markides KS, Levin JS, Ray LA: Determinants of physician utilization among Mexican Americans: a three-generation study. Med Care 1985; 23:236-246.

AJPH June 1987, Vol. 77, No. 6

SMOKING AMONG MEXICAN AMERICANS 16. Markides KS, Vernon SW: Aging, sex-role orientation and adjustment. J Gerontol 1984; 39:586-591. 17. Cuellar I, Harris LC, Jasso R: An acculturation scale for Mexican American normal and clinical populations. Hispanic J Behav Sci 1980; 2:199. 18. Cox DR: The analysis of binary data. London: Methuen, 1970.

19. Sunseri AJ, Alberti JM, Kend ND, et al: Reading, demographic, social and psychological factors related to pre-adolescent smoking and non-smoking behaviors and attitudes. JOSH 1983; 53:257-263. 20. Hunter SM, Baugh JG, Webber LS, et al: Social learning effects on trial and adoption of cigarette smoking in children: the Bogalusa Heart Study. Prev Med 1982; 11:2942.

Survey Shows Most Localities Have 'No Smoking' or 'Restricted' Policies Among 65 city, county and district health departments surveyed recently, 72 to 83 per cent have policies which limit smoking-either by local or state policies. State policies often take the form of state clean indoor air legislation. The survey, conducted by the US Conference of Local Health Officers (an affiliate of the US Conference of Mayors) found that the policies covered "no smoking" or "designated smoking" in schools, public facilities, worksites, or restaurants. Half of the communities surveyed have issued voluntary guidelines covering smoking, i.e., some localities request that restaurants above a certain seating capacity designate a percentage of space "no smoking". The survey also showed that three out of four participating health departments operate smoking education programs targeted to specific groups, such as pregnant women, school children, health department employees and clients, teachers, and private businesses. Eighty-seven per cent of the departments are involved in coordinated smoking education programs with the American Cancer Society, American Heart Association, and American Lung Association. Among the anti-smoking initiative reported in the survey: * Austin, Texas restricts smoking on worksites and requires employers to develop written smoking policies; * Houston, Texas requires employers of 20 or more workers to develop smoking policies; * Cincinnati, Ohio restricts smoking in commercial establishments and other open landscaped office areas; * Kansas City, Missouri restricts smoking in public spaces and certain workplaces, including commercial establishments; Copies of the survey and additional information on anti-smoking initiatives are available from Alan Gambrell, US Conference of Local Health Officers, 1620 Eye Street, NW, Washington, DC 20006. Tel: 202/293-7330. In addition, the US Conference of Mayors will hold a related workshop for registrants at its annual meeting June 13-17, 1987 in Nashville, Tennessee. The workshop will focus on such activities as no smoking legislative initiatives, model legislation, and enforcement, and address the range of ordinances developed in various regions of the country.

AJPH June 1987, Vol. 77, No. 6

711