- Email: [email protected]
Prevalence and correlates of breast and cervical cancer screening among older women
Prevalence and Correlates of Breast and Cervical Cancer Screening Among Older Women HIRSCH
S. RUCHLIN,
PhD
Objective: To identify and assess differences in cancer screening patterns among women 55-64, 65-74, 75-84, and over 84 years of age. Methods: Nationally representative data reported in the 1990 Health Promotion and Disease Prevention Supplement to the National Health Interview Survey of 28,584,574 women were analyzed secondarily. The dependent variables were a knowledge of breast self-examination, ever having had a mammogram, and a Papanicolaou smear within the last 3 years. Independent variables were age and various sociodemographic, health-status, and health-belief measures. Results: More than half (58%) of the women had ever had a mammogram, and of these, 91% had had between one and five mammograms. Over a third (35%) of those who had not had a mammogram attributed the omission to a lack of a recommendation by a physician. Almost half (45%) had had a breast examination by a physician within the last year, and 84% knew how to examine their own breasts. Approximately 87% had a Papanicolaou smear within the last 3 years. Age, race, education, and living in a large city were significantly associated with all three screening measures, but prevalent health beliefs were significantly associated only with breastcancer screening. Conclusion: Lack of mammogram screening in a substantial number of women, attributed to lack of physician recommendation, decreased screening in the older age groups, and the negative association of all three screening tests with race and residence in a large city suggest that new interventions are needed by health care providers and the public health community to increase older women’s use of effective cancer screening techniques. (Obstet Gynecoll997; 90~16-21. 0 1997 by The American College of Obstetricians and Gynecologists.)
Health promotion and disease prevention are national health priorities. Specific targets and guidelines have From the Departments of Public Health md Medicine, Cortwll University Med:cal College, &zcl York, New York; arzd the Program in Clinical Eviderumlow and Health Serzlices Research. Cornell Unizwrsitu Graduate ‘School of‘iiiedical Sciences, Neul York, Neul York. a This research zbas su pported by a grarft from the AARP Andrus Foundation.
16
002Y-7844/97/$17.00 1’11SOO2Y-7844(Y7)00220-2
been issued with regard to the use of breast selfexamination, mammography, and Papanicolaou smears in women over the age of 50 years.l,’ The U.S. Preventive Services Task Force recommends routine screening for breast and cervical cancer every l-2 years with mammography alone or mammography and annual breast examinations for women aged 50-69. Being considered high risk may justify such screening above age 69.3 The ACOG recommends annual mammography in women over the age of 50.* These guidelines are supported by evidence that early screening can reduce the rate of mortality from breast and cervical cancer.5-7 Piani and Schoenborn8 found that 30% of women 65 years of age or older had Papanicolaou smearsand 42% had breast examinations by a health professional in 1990. Eighty-one percent knew how to do breast selfexamination, but only 45% did so in that year. Data on mammography was reported, and the percentage, who had “ever” had a mammogram was 6862, and 50% for women 50-59, 60-69 and 70 or more years of age, respectively, while the percentage who had had a mammogram in the past 3 years was 60,55, and 42% in the three age groups, respectively. All these rates are below those recommended by government and medical groups. The existing knowledge base on this aspect of health promotion and diseaseprevention for older women is quite limited. National data provide information for persons “65 and older,” despite recognition in the gerontologic community of significant differences in capacities and needs of diverse populations of older adults. Gerontologists prefer to study the older population using more refined age categories such as the “young old” (55-74), the “old” (75-84), and the “very old” (85 and over).” Furthermore, little attention has been given to identifying the factors that are associated with differences in cancer screening levels among older women. The objectives of this study were to generate detailed national point-prevalence profiles for cancer screening among older women within the 55-64, 65-74, 75-84,
Obstetrics 6 Gynecology
and over 84 age categories, and to investigate the impact of age and other sociodemographic, healthstatus, and health-belief factors on three measures of cancer screening: a knowledge of breast selfexamination, ever having had a mammogram, and a Papanicolaou smear within the last 3 years.
Materials and Methods Data for this study were derived from the 1990 Health Promotion and Disease Prevention Supplement to the National Health Interview Survey. The National Health Interview Survey is a continuous, nationwide probability survey of the civilian noninstitutionalized population of the United States. Weights are provided to project sample data to national levels. One adult per family was randomly selected to respond to the questions in the Health Promotion and Disease Prevention Supplement, which included information on a woman’s use of cancer screening methods, sociodemographic characteristics, self-reported health status, and attitudes on some activities that are associated with adverse health effects, such as smoking, drinking, and being overweight. Two separate analyses were conducted. The first developed national point-prevalence estimates by age category for the following nine cancer screening activities: ever having had a mammogram, number of mammograms, reason for the mammogram, the most important reason for no recent mammogram, plan to have a mammogram in the future, time since last breast examination by a physician, knowledge of how to examine own breasts, times per year she examines own breasts, and time since her last Papanicolaou smear. The second analysis identified, through the use of multivariate logistic regression analysis, the extent to which sociodemographic factors, health status, and health beliefs are associated with the likelihood of a woman knowing how to examine her own breasts, ever having a mammogram, and having a Papanicolaou smear within the last 3 years. The independent variables used in the multiple logistic regression analysis were age, sex, race, educational attainment, living in a central city of a metropolitan statistical area, labor-force participation, geographic location, self-assessed health status, having diabetes, ever being hypertensive, being overweight, and responding to all questions without the assistance of a proxy respondent (a surrogate measure for being cognitively intact). Also used were the belief that smoking increases the chance of developing heart disease, lung cancer, and emphysema; that heavy drinking increases the chance of getting cirrhosis of the liver; and that being overweight increases the chance of getting heart disease.
VOL.
90, NO.
1, JULY
1997
Since the Health Promotion and Disease Prevention Survey did not have any direct health-belief questions associated with the efficacy of cancer screening, these health-belief questions were used as proxies for whether a respondent subscribed to commonly held health beliefs. Numerous other independent variables were available in the Health Promotion and Disease Prevention Survey data set and were considered candidates for use in this study. Concern for multicolinearity dictated that not all of the sociodemographic, health-status, and health-belief variables reported in the data set be used simultaneously in any regression run. Data plots and correlation analyses were used as aids in suggesting the variables that were selected for use. Following the methodology used by the U.S. Congress’ Office of Technology Assessment,” a correlation coefficient greater than 0.25 or less than -0.25 was used to infer multicolinearity. With the exception of the health-belief variable group, such a correlation existed for only one pair of variables, being in the labor force and being in age category 75-84 (r = -0.26). Correlation coefficients ranging between 0.30 and 0.48 existed within the group of health-belief variables. This was not unexpected because a person who subscribes to one of the common health beliefs probably subscribes to most of them. To note the impact of the various health-belief variables, all were retained, and, in the regression equations, many still achieved statistical significance. One should remember that these significant values are biased in a downward direction, because of the existent multicolinearity. Weighted cross-tabulations and regression equations were performed with the Software for Survey Data Analysis (Research Triangle Institute, Research Triangle Park, NC). This program uses Taylor series linearization to calculate variances that take into account the complex, multistage sampling inherent in the National Health Interview Survey. Chi-square statistics were used to assess whether statistically significant pointprevalence differences existed across age groups. Adjusted odds ratios (ORs), calculated through the use of logistic regression equations, were used to investigate the importance of each specific independent variable (the term adjusted implies that each OR was calculated with the impact of all the other independent variables in the equation held constant). When multiple values existed for any variable (ie, age, geographic location, and health status), one category was designated as the reference category and omitted from the regression equation. Confidence intervals were calculated for each OR, using a probability value of .05; ORs whose confidence intervals do not include the value of 1 are significant at P = .05 or lower.
Ruchlin
Cancer
Screening
in Older
Women
17
Table
1.
Characteristics of the Studv Samule Variable
n
%
5564 (reference category) 65-74 75-84 X4 White Education (1 or more years of college) Live in central city of a metropolitan statistical area Participate in labor force Geographic location Live in Northeast (reference category) Live in Midwest Live in South Live in West Health status Excellent or very good (reference category) Good Fair or poor Have diabetes Ever had hypertension Overweight Responded self entirely Believe smoking definitely increases chance of developing heart disease Believe smoking definitely increases chance of developing lung cancer Believe smoking definitely increases chance of developing emphysema Believe heavy drinking increases chance of getting cirrhosis of the liver Believe being very overweight definitely increases chance of heart disease
11,217,753 10,018,511 5,895,579 1,452,731 25,413,OOO 6,348,810
39.2 35.0 20.6 5.1 88.9 22.3
21,209,754
74.2
7,203,313
25.2
6,378,479
22.3
7,067,753 9,906,412 5,231,930
24.7 34.7 18.3
12,235,080
42.9
9,638,405 6,651,231 2,417,229 12,437,850 14,295,260 24,790,960 18,933,513
33.8 23.3 8.6 44.0 50.3 87.2 72.6
22,850,460
84.6
22,231,774
83.4
22,971,672
86.2
19,873,638
73.8
Age (Y)
Data reflect each variable.
the percentage
of the sample
providing
responses
for
Red ts The sociodemographic, health-status, and health-belief characteristics of the 28,584,574 women above the age of 54 in the sample are shown in Table 1. Most women had had a mammogram during their lifetime, although at older ages fewer women reported ever having had this screening test (P < .Ol) (Table 2). For those who did have a mammogram, almost all indicated that they had had between one and five mammograms, a rate that increased with age (P < .Ol). The modal reason given for having a mammogram was that it was part of a routine checkup, and this response remained fairly constant with age. Slightly more than a third of the women who did not have a recent mammogram noted that it was not
18
Ruchlin
Cancer
Screening
in Older
Women
recommended by a physician, and over a quarter did not think it was necessary. A significant age-group pattern (P < .Ol) emerged for this question: At higher ages, more women gave this response. More than half of the respondents said that they plan to have a mammogram in the future. About a quarter indicated that they did not plan to have a mammogram, and slightly more than a fifth said they would have one upon a physician recommendation. At higher ages, fewer women indicated an intent to have a mammogram in the future (P < .Ol). Almost half of the women reported that they had had a breast examination by a physician within the survey year, and the next most common response was within 3 years. At older ages, the interval between breast examinations increased (P < .Ol). Most women knew how to examine their own breasts, but at higher ages, knowledge of this procedure declined (P < .Ol). The modal response to the question regarding the number of times per year a woman examined her own breasts was in the range of one to 12 times per year. At higher ages, fewer women examined their own breasts at the modal frequency. Most women reported that they had had a Papanicolaou smear within the last 3 years, although at older ages there was a longer time interval between tests (P < .Ol). The correlates of screening are shown in Table 3. Women older than 64 displayed lower odds of having each cancer screening test than women in the 55-64 age group. Furthermore, the ORs decreased with higher age groups. White women had higher likelihood of knowing how to examine their own breasts and ever having had a mammogram than did nonwhite women. Having attended college and living in a central city of a metropolitan statistical area raised the odds for all three screening activities. Being in the labor force had no significant association with breast and cervical cancer screening. The impact of geographic region varied across screening activities. Women living in the Midwest had higher odds of knowing how to examine their own breasts relative to those living in the Northeast. Those living in the West had higher odds of ever having had a mammogram, and those living in the South and West had higher odds of having had a Papanicolaou smear than did older women living in the Northeast. Reporting one’s health status as fair or poor rather than excellent or very good lowered the odds of a woman knowing how to examine her own breasts. Ever having had hypertension raised the odds of a woman’s ever having had a mammogram. Believing that smoking definitely increases the chance of developing heart disease raised the odds of a woman knowing how to examine her own breasts and having a mammogram, as
Obstefrics
6 Gynecology
Table
2. Female
Health
Habits
By
Age Category,
1990 Age (Y)
Health
habit
Ever had a mammogram Yes No Number of mammograms 1-5 6-10 >10 Reason for the mammogram Routine checkup Family history Breast symptom or condition Other reasons Most important reason why no recent mammogram No problems Not recommended by physician Didn’t think it was necessary Financial reasons Other reasons Plan to have a mammogram in the future Yes If physician recommends No Time since last breast exam by a physician (y) <1 1-3 4-10 >lO Know how to examine own breasts Yes No Times per year examine own breasts Never or48 Time since last I’apanicolaou smear (y) 1-3 4-10 >lO
Total (%)
1997
(%)
75-84
(%)
285 (%)
58.7 41.3
49.7 50.3
30.1 69.9
91 .o 6.6 2.4
88.8 8.3 2.9
90.6 6.8 2.6
94.6 4.1 1.3
96.7 2.5 0.8
82.1 0.9 15.3 1.7
81.3 0.9 16.4 1.4
83.2 1.2 13.6 2.0
82.2 0.6 15.3 1.9
80.7 0 17.6 1.7
15.6 34.7 26.7 4.9 18.1
16.2 29.4 22.6 8.6 23.2
15.4 34.8 28.0 4.1 17.7
14.8 40.0 29.6 1.9 13.7
17.2 41.5 30.0 0.5 10.8
53.0 22.4 24.6
65.6 18.4 16.0
54.2 22.4 23.4
34.8 27.9 37.3
15.0 32.1 52.9
44.5 33.2 10.3 12.0
48.2 33.4 10.2 8.2
43.6 34.6 10.6 11.2
40.7 32.6 9.6 17.1
36.6 23.4 11.9 28.1
84.1 15.9
89.7 10.3
85.2 14.8
76.5 23.5
64.0 36.0
17.0 63.3 3.2 1.6 14.9
14.4 65.6 3.2 1.7 15.1
15.7 64.5 3.2 1.8 14.8
22.4 57.5 3.7 1.2 15.2
34.5 52.7 1.6 0 11.2
86.7 7.5 5.8
90.8 4.7 4.5
86.7 7.4 5.9
81.3 11.3 7.4
76.6 15.3 8.1
The point-prevalence data reported here indicates the heterogeneity of the older adult cohort and the need to report data by smaller age intervals than the 65 years of age and over category that is commonly used in healthservices research. Sociodemographic, health-status, and health-belief variables are associated with older women’s use of cancer screening procedures. At older ages, women display a reduced use of these life-saving techniques. White women display higher utilization
1, JULY
65-74
65.2 34.8
Discussion
90, NO.
(%)
58.1 41.9
did the belief that smoking definitely increases the chance of developing emphysema.
VOL.
55-64
levels than nonwhite women, possibly because of access, a factor not directly addressed in this study. Women who attended at least 1 year of college had higher odds of engaging in these health-enhancement techniques. Reporting one’s health status as fair or poor rather than excellent or very good was associatedwith lower odds of knowing how to examine one’s breasts. As hypothesized, “positive” health beliefs were associated with greater use of breast cancer screening procedures. The reason that they were not related to the use of Papanicolaou smears merits further research. Findings similar to those reported here for mammography have been reported by other researchers. Mor et a1,5 analyzing data from the 1987 National Health
Ruchlin
Cancer Screening in Older Women
19
Table
3. Multivariate
Correlates
of Screening
for
Cancer Woman knows
Variable
to examine
Age 65-74 Age 75-84 Age over 84 White 1 or more years of college Lives in central city of a metropolitan statistical area Participates in labor force Lives in Midwest Lives in South Lives in West Health status good Health status fair or poor Have diabetes Ever had hypertension Overweight Respondent answered all questions Believes smoking definitely increases chance of developing heart disease Believes smoking definitely increases chance of developing lung cancer Believes smoking definitely increases chance of developing emphysema Believes heavy drinking increases chance of getting cirrhosis of the liver Believes being very overweight definitely increases chance of heart disease
how breasts
Woman ever had a mammogram
Papanicolaou smear within last 3 years
0.73 0.41 0.29 1.90 1.57 1.23
(0.50,0.97) (0.16, 0.67) (0.00, 0.68) (1.67, 2.14) (1.35, 1.79) (1.01, 1.45)
0.76 0.57 0.22 1.30 1.75 1.28
(0.58,0.94) (0.39,0.75) (0.00,0.55) (1.08, 1.52) (1.61, 1.89) (1.10, 1.46)
0.64 0.46 0.33 1.00 1.57 1.21
(0.42,0.86) (0.21,0.72) (0.00,0.70) (0.69, 1.31) (1.35, 1.79)* (1.03, 1.39)
1.09 1.45 1.12 1.25 0.79 0.72 1.08 1.14 1.15 1.06 1.48
(0.86, 1.33) (1.14, 1.76) (0.89, 1.36) (0.97,1.52) (0.57, 1.01) (0.47, 0.98) (0.77, 1.39) (0.96, 1.32) (0.95, 1.35) (0.75, 1.37) (1.26, 1.70)
1.05 0.85 0.80 1.27 0.99 0.88 0.96 1.23 1.02 0.96 1.31
(0.89, 1.21) (0.65, 1.05) (0.60, 1.00) (1.07,1.47) (0.85, 1.13) (0.72, 1.04) (0.74, 1.18) (1.11, 1.35) (0.90, 1.14) (0.76, 1.16) (1.15, 1.47)
0.95 1.17 1.34 1.28 1.07 0.87 0.98 1.13 1.13 0.70 1.04
(0.72, 1.19) (0.94, 1.41) (1.11, 1.58) (1.03, 1.54) (0.87, 1.27) (0.64,l.ll) (0.67, 1.29) (0.95,1.31) (0.95, 1.31) (0.39,l.Ol) (0.82,1.26)
0.91 (0.60, 1.22)
1.12 (0.90,1.34)
1.03 (0.76, 1.30)
1.38 (1.11,1.65)
1.34 (1.12, 1.56)
1.13 (0.86,1.40)
1.13 (0.88, 1.39)
1.17 (0.97, 1.37)
0.90 (0.65, 1.16)
1.52 (1.30, 1.74)
1.08 (0.94, 1.22)
1.07 (0.85,1.29)
Data are presented as odds ratio (95% confidence interval). * Odds ratios are statistically significant at P < .05
Interview Survey and the 1984-1988 Longitudinal Study of Aging, noted that after controlling for sociodemographic factors, health status was unrelated to having a mammogram. Urban and colleagues,” reporting on the results of a 1989 telephone survey of 1538 women between the ages of 50 and 75, noted that educational attainment was not significantly related to having a mammogram within the last 2 years. Analyzing 1991 Medicare part B data for women above the age of 64 living in ten states across the country, Burns et all* noted that women in the age categories 75-84 and 84 and older reported a lower use of mammography than did those in the 6574 age group. They also reported lower use rates for nonwhite women. The important role played by physicians in the use of these screening tests is directly and indirectly noted in this study. Data reported in Table 2 under the category “Most important reason why no recent mammogram” indicate that over a third of the women noting the absence of this preventive technique claimed that it was not recommended by a physician. This finding and others reported in the literaturer3~i6 indicate that physicians who perceive themselves as rendering primary care to women need to be advocates for such important
20
Ruchlin
Cancer
Screening
in Older
Women
life-saving procedures. Medicare’s recent coverage of mammography and periodic screening for cervical cancer, should facilitate older women’s access to these important procedures, even though only a small number of respondents noted that financial reasons deterred them from having a mammogram (Table 2). If such coverage is not routinely available to women younger than 65, major efforts should be launched to remedy this deficit. Two limitations regarding the data set used in this study merit attention. First, all data in the Health Promotion and Disease Prevention Surveys are selfreported. However, for population surveillance of mammography, self-reported data are generally valid.17 Second, the data set does not permit one to recognize or control for differential mortality due to breast or cervical cancer, a phenomenon common to most existing data sets for all age groups.
References 1. US Department 2000: National
of Health and Human Services. Healthy health promotion and disease prevention
Obstetrics
people objec-
6 Gynecology
2. 3. 4.
5. 6.
7.
8.
9.
10.
11.
12.
VOL.
tives. DHHS publication no. (PHS) 91-50212. Washington, DC: US Government Printing Office, 1991. Special Medical Reports. Recommendations from AHCPR mammography guideline. Am Fam Phys 1994;50:182~5. Report of the US Preventive Services Task Force. Guide to clinical preventive services. Baltimore: Williams & Wilkins, 199673-87. American College of Obstetricians and Gynecologists. Carcinoma of the breast. ACOG technical bulletin No. 158. Washington, DC: American College of Obstetricians and Gynecologists, 1991. Mor V, Pacala JT, Rakowski W. Mammography for older women: Who uses, who benefits? J Gerontol 1992;47:43-9. Kerlikowske K, Grady D, Rubin SM, Sandrock C, Emester VL. Efficacy of screening mammography: A meta-analysis. JAMA 1995;273:149-54. Muller C, Mandelblatt J, Schechter CB. Costs and effectiveness of cervical cancer screening in elderly women. OTA-BP-H-65. Washington, DC: US Government Printing Office, 1990. Piani A, Schoenbom C. Health promotion and disease prevention: United States. 1990. DHHS Publication No. (PHS) 93-1513, Series 10, No. 185. Washington, DC: US Government Printing Office, 1993. Maddox GL. Intervention strategies to enhance well-being in later life: The status and prospect of guided change. Health Serv Res 1985;19:1007-18. Office of Technology Assessment. The use of preventive services by the elderly. Preventive health services under Medicare, paper #2. Washington, DC: US Government Printing Office, 1989. Urban N, Anderson GL, Peacock S. Mammography screening: How important is cost as a barrier to use? Am J Public Health 1994;84:50-5. Bums RB, McCarthy EP, Freund KM, Marwill SL, Shwartz M, Ash A, et al. Variability in mammography use among older women. J Am Geriatr Sot 1996;44:922-6.
90, NO.
1, JULY
1997
I3 14.
15.
16. 17.
Mandelblatt JS, Phillips RN. Cervical cancer: How often-and why-to screen older women. Geriatrics 1996;51:45-8. Breen N, Kessler L. Changes in the use of screening mammography: Evidence from the 1987 and 1990 National Health Interview Surveys. Am J Public Health 1994;84:62-7. Skinner CS, Strecher VJ, Hospers H. Physicians’ recommendations for mammography: Do tailored messages make a difference? Am J Public Health 1994;84:43-9. Kruse J, Phillips DM. Factors influencing women’s decision to undergo mammography. Obstet Gynecol 1987;70:744-7. Zapka JG, Bigelow C, Hurley T, Ford LD, Egelhofer J, Cloud WM, et al. Mammography use among sociodemogaphically diverse women: The accuracy of self-report. Am J Public Health 1996;86: 1016-21.
Address
reprint
requests
Hirsch Cornell
to:
S. Rucklin, PkD University Medical 2300 York Avenue, Box 4 New York, NY 10021
College
Received December 27, 1996. Received in revised form March Accepted March 18, 1997.
13, 1997
Copyright 0 1997 by The American College of Obstetricians Gynecologists. Published by Elsevier Science Inc.
Ruchlin
Cancer
Screening
in Older
Women
and
21
S. RUCHLIN,
PhD
Objective: To identify and assess differences in cancer screening patterns among women 55-64, 65-74, 75-84, and over 84 years of age. Methods: Nationally representative data reported in the 1990 Health Promotion and Disease Prevention Supplement to the National Health Interview Survey of 28,584,574 women were analyzed secondarily. The dependent variables were a knowledge of breast self-examination, ever having had a mammogram, and a Papanicolaou smear within the last 3 years. Independent variables were age and various sociodemographic, health-status, and health-belief measures. Results: More than half (58%) of the women had ever had a mammogram, and of these, 91% had had between one and five mammograms. Over a third (35%) of those who had not had a mammogram attributed the omission to a lack of a recommendation by a physician. Almost half (45%) had had a breast examination by a physician within the last year, and 84% knew how to examine their own breasts. Approximately 87% had a Papanicolaou smear within the last 3 years. Age, race, education, and living in a large city were significantly associated with all three screening measures, but prevalent health beliefs were significantly associated only with breastcancer screening. Conclusion: Lack of mammogram screening in a substantial number of women, attributed to lack of physician recommendation, decreased screening in the older age groups, and the negative association of all three screening tests with race and residence in a large city suggest that new interventions are needed by health care providers and the public health community to increase older women’s use of effective cancer screening techniques. (Obstet Gynecoll997; 90~16-21. 0 1997 by The American College of Obstetricians and Gynecologists.)
Health promotion and disease prevention are national health priorities. Specific targets and guidelines have From the Departments of Public Health md Medicine, Cortwll University Med:cal College, &zcl York, New York; arzd the Program in Clinical Eviderumlow and Health Serzlices Research. Cornell Unizwrsitu Graduate ‘School of‘iiiedical Sciences, Neul York, Neul York. a This research zbas su pported by a grarft from the AARP Andrus Foundation.
16
002Y-7844/97/$17.00 1’11SOO2Y-7844(Y7)00220-2
been issued with regard to the use of breast selfexamination, mammography, and Papanicolaou smears in women over the age of 50 years.l,’ The U.S. Preventive Services Task Force recommends routine screening for breast and cervical cancer every l-2 years with mammography alone or mammography and annual breast examinations for women aged 50-69. Being considered high risk may justify such screening above age 69.3 The ACOG recommends annual mammography in women over the age of 50.* These guidelines are supported by evidence that early screening can reduce the rate of mortality from breast and cervical cancer.5-7 Piani and Schoenborn8 found that 30% of women 65 years of age or older had Papanicolaou smearsand 42% had breast examinations by a health professional in 1990. Eighty-one percent knew how to do breast selfexamination, but only 45% did so in that year. Data on mammography was reported, and the percentage, who had “ever” had a mammogram was 6862, and 50% for women 50-59, 60-69 and 70 or more years of age, respectively, while the percentage who had had a mammogram in the past 3 years was 60,55, and 42% in the three age groups, respectively. All these rates are below those recommended by government and medical groups. The existing knowledge base on this aspect of health promotion and diseaseprevention for older women is quite limited. National data provide information for persons “65 and older,” despite recognition in the gerontologic community of significant differences in capacities and needs of diverse populations of older adults. Gerontologists prefer to study the older population using more refined age categories such as the “young old” (55-74), the “old” (75-84), and the “very old” (85 and over).” Furthermore, little attention has been given to identifying the factors that are associated with differences in cancer screening levels among older women. The objectives of this study were to generate detailed national point-prevalence profiles for cancer screening among older women within the 55-64, 65-74, 75-84,
Obstetrics 6 Gynecology
and over 84 age categories, and to investigate the impact of age and other sociodemographic, healthstatus, and health-belief factors on three measures of cancer screening: a knowledge of breast selfexamination, ever having had a mammogram, and a Papanicolaou smear within the last 3 years.
Materials and Methods Data for this study were derived from the 1990 Health Promotion and Disease Prevention Supplement to the National Health Interview Survey. The National Health Interview Survey is a continuous, nationwide probability survey of the civilian noninstitutionalized population of the United States. Weights are provided to project sample data to national levels. One adult per family was randomly selected to respond to the questions in the Health Promotion and Disease Prevention Supplement, which included information on a woman’s use of cancer screening methods, sociodemographic characteristics, self-reported health status, and attitudes on some activities that are associated with adverse health effects, such as smoking, drinking, and being overweight. Two separate analyses were conducted. The first developed national point-prevalence estimates by age category for the following nine cancer screening activities: ever having had a mammogram, number of mammograms, reason for the mammogram, the most important reason for no recent mammogram, plan to have a mammogram in the future, time since last breast examination by a physician, knowledge of how to examine own breasts, times per year she examines own breasts, and time since her last Papanicolaou smear. The second analysis identified, through the use of multivariate logistic regression analysis, the extent to which sociodemographic factors, health status, and health beliefs are associated with the likelihood of a woman knowing how to examine her own breasts, ever having a mammogram, and having a Papanicolaou smear within the last 3 years. The independent variables used in the multiple logistic regression analysis were age, sex, race, educational attainment, living in a central city of a metropolitan statistical area, labor-force participation, geographic location, self-assessed health status, having diabetes, ever being hypertensive, being overweight, and responding to all questions without the assistance of a proxy respondent (a surrogate measure for being cognitively intact). Also used were the belief that smoking increases the chance of developing heart disease, lung cancer, and emphysema; that heavy drinking increases the chance of getting cirrhosis of the liver; and that being overweight increases the chance of getting heart disease.
VOL.
90, NO.
1, JULY
1997
Since the Health Promotion and Disease Prevention Survey did not have any direct health-belief questions associated with the efficacy of cancer screening, these health-belief questions were used as proxies for whether a respondent subscribed to commonly held health beliefs. Numerous other independent variables were available in the Health Promotion and Disease Prevention Survey data set and were considered candidates for use in this study. Concern for multicolinearity dictated that not all of the sociodemographic, health-status, and health-belief variables reported in the data set be used simultaneously in any regression run. Data plots and correlation analyses were used as aids in suggesting the variables that were selected for use. Following the methodology used by the U.S. Congress’ Office of Technology Assessment,” a correlation coefficient greater than 0.25 or less than -0.25 was used to infer multicolinearity. With the exception of the health-belief variable group, such a correlation existed for only one pair of variables, being in the labor force and being in age category 75-84 (r = -0.26). Correlation coefficients ranging between 0.30 and 0.48 existed within the group of health-belief variables. This was not unexpected because a person who subscribes to one of the common health beliefs probably subscribes to most of them. To note the impact of the various health-belief variables, all were retained, and, in the regression equations, many still achieved statistical significance. One should remember that these significant values are biased in a downward direction, because of the existent multicolinearity. Weighted cross-tabulations and regression equations were performed with the Software for Survey Data Analysis (Research Triangle Institute, Research Triangle Park, NC). This program uses Taylor series linearization to calculate variances that take into account the complex, multistage sampling inherent in the National Health Interview Survey. Chi-square statistics were used to assess whether statistically significant pointprevalence differences existed across age groups. Adjusted odds ratios (ORs), calculated through the use of logistic regression equations, were used to investigate the importance of each specific independent variable (the term adjusted implies that each OR was calculated with the impact of all the other independent variables in the equation held constant). When multiple values existed for any variable (ie, age, geographic location, and health status), one category was designated as the reference category and omitted from the regression equation. Confidence intervals were calculated for each OR, using a probability value of .05; ORs whose confidence intervals do not include the value of 1 are significant at P = .05 or lower.
Ruchlin
Cancer
Screening
in Older
Women
17
Table
1.
Characteristics of the Studv Samule Variable
n
%
5564 (reference category) 65-74 75-84 X4 White Education (1 or more years of college) Live in central city of a metropolitan statistical area Participate in labor force Geographic location Live in Northeast (reference category) Live in Midwest Live in South Live in West Health status Excellent or very good (reference category) Good Fair or poor Have diabetes Ever had hypertension Overweight Responded self entirely Believe smoking definitely increases chance of developing heart disease Believe smoking definitely increases chance of developing lung cancer Believe smoking definitely increases chance of developing emphysema Believe heavy drinking increases chance of getting cirrhosis of the liver Believe being very overweight definitely increases chance of heart disease
11,217,753 10,018,511 5,895,579 1,452,731 25,413,OOO 6,348,810
39.2 35.0 20.6 5.1 88.9 22.3
21,209,754
74.2
7,203,313
25.2
6,378,479
22.3
7,067,753 9,906,412 5,231,930
24.7 34.7 18.3
12,235,080
42.9
9,638,405 6,651,231 2,417,229 12,437,850 14,295,260 24,790,960 18,933,513
33.8 23.3 8.6 44.0 50.3 87.2 72.6
22,850,460
84.6
22,231,774
83.4
22,971,672
86.2
19,873,638
73.8
Age (Y)
Data reflect each variable.
the percentage
of the sample
providing
responses
for
Red ts The sociodemographic, health-status, and health-belief characteristics of the 28,584,574 women above the age of 54 in the sample are shown in Table 1. Most women had had a mammogram during their lifetime, although at older ages fewer women reported ever having had this screening test (P < .Ol) (Table 2). For those who did have a mammogram, almost all indicated that they had had between one and five mammograms, a rate that increased with age (P < .Ol). The modal reason given for having a mammogram was that it was part of a routine checkup, and this response remained fairly constant with age. Slightly more than a third of the women who did not have a recent mammogram noted that it was not
18
Ruchlin
Cancer
Screening
in Older
Women
recommended by a physician, and over a quarter did not think it was necessary. A significant age-group pattern (P < .Ol) emerged for this question: At higher ages, more women gave this response. More than half of the respondents said that they plan to have a mammogram in the future. About a quarter indicated that they did not plan to have a mammogram, and slightly more than a fifth said they would have one upon a physician recommendation. At higher ages, fewer women indicated an intent to have a mammogram in the future (P < .Ol). Almost half of the women reported that they had had a breast examination by a physician within the survey year, and the next most common response was within 3 years. At older ages, the interval between breast examinations increased (P < .Ol). Most women knew how to examine their own breasts, but at higher ages, knowledge of this procedure declined (P < .Ol). The modal response to the question regarding the number of times per year a woman examined her own breasts was in the range of one to 12 times per year. At higher ages, fewer women examined their own breasts at the modal frequency. Most women reported that they had had a Papanicolaou smear within the last 3 years, although at older ages there was a longer time interval between tests (P < .Ol). The correlates of screening are shown in Table 3. Women older than 64 displayed lower odds of having each cancer screening test than women in the 55-64 age group. Furthermore, the ORs decreased with higher age groups. White women had higher likelihood of knowing how to examine their own breasts and ever having had a mammogram than did nonwhite women. Having attended college and living in a central city of a metropolitan statistical area raised the odds for all three screening activities. Being in the labor force had no significant association with breast and cervical cancer screening. The impact of geographic region varied across screening activities. Women living in the Midwest had higher odds of knowing how to examine their own breasts relative to those living in the Northeast. Those living in the West had higher odds of ever having had a mammogram, and those living in the South and West had higher odds of having had a Papanicolaou smear than did older women living in the Northeast. Reporting one’s health status as fair or poor rather than excellent or very good lowered the odds of a woman knowing how to examine her own breasts. Ever having had hypertension raised the odds of a woman’s ever having had a mammogram. Believing that smoking definitely increases the chance of developing heart disease raised the odds of a woman knowing how to examine her own breasts and having a mammogram, as
Obstefrics
6 Gynecology
Table
2. Female
Health
Habits
By
Age Category,
1990 Age (Y)
Health
habit
Ever had a mammogram Yes No Number of mammograms 1-5 6-10 >10 Reason for the mammogram Routine checkup Family history Breast symptom or condition Other reasons Most important reason why no recent mammogram No problems Not recommended by physician Didn’t think it was necessary Financial reasons Other reasons Plan to have a mammogram in the future Yes If physician recommends No Time since last breast exam by a physician (y) <1 1-3 4-10 >lO Know how to examine own breasts Yes No Times per year examine own breasts Never or
Total (%)
1997
(%)
75-84
(%)
285 (%)
58.7 41.3
49.7 50.3
30.1 69.9
91 .o 6.6 2.4
88.8 8.3 2.9
90.6 6.8 2.6
94.6 4.1 1.3
96.7 2.5 0.8
82.1 0.9 15.3 1.7
81.3 0.9 16.4 1.4
83.2 1.2 13.6 2.0
82.2 0.6 15.3 1.9
80.7 0 17.6 1.7
15.6 34.7 26.7 4.9 18.1
16.2 29.4 22.6 8.6 23.2
15.4 34.8 28.0 4.1 17.7
14.8 40.0 29.6 1.9 13.7
17.2 41.5 30.0 0.5 10.8
53.0 22.4 24.6
65.6 18.4 16.0
54.2 22.4 23.4
34.8 27.9 37.3
15.0 32.1 52.9
44.5 33.2 10.3 12.0
48.2 33.4 10.2 8.2
43.6 34.6 10.6 11.2
40.7 32.6 9.6 17.1
36.6 23.4 11.9 28.1
84.1 15.9
89.7 10.3
85.2 14.8
76.5 23.5
64.0 36.0
17.0 63.3 3.2 1.6 14.9
14.4 65.6 3.2 1.7 15.1
15.7 64.5 3.2 1.8 14.8
22.4 57.5 3.7 1.2 15.2
34.5 52.7 1.6 0 11.2
86.7 7.5 5.8
90.8 4.7 4.5
86.7 7.4 5.9
81.3 11.3 7.4
76.6 15.3 8.1
The point-prevalence data reported here indicates the heterogeneity of the older adult cohort and the need to report data by smaller age intervals than the 65 years of age and over category that is commonly used in healthservices research. Sociodemographic, health-status, and health-belief variables are associated with older women’s use of cancer screening procedures. At older ages, women display a reduced use of these life-saving techniques. White women display higher utilization
1, JULY
65-74
65.2 34.8
Discussion
90, NO.
(%)
58.1 41.9
did the belief that smoking definitely increases the chance of developing emphysema.
VOL.
55-64
levels than nonwhite women, possibly because of access, a factor not directly addressed in this study. Women who attended at least 1 year of college had higher odds of engaging in these health-enhancement techniques. Reporting one’s health status as fair or poor rather than excellent or very good was associatedwith lower odds of knowing how to examine one’s breasts. As hypothesized, “positive” health beliefs were associated with greater use of breast cancer screening procedures. The reason that they were not related to the use of Papanicolaou smears merits further research. Findings similar to those reported here for mammography have been reported by other researchers. Mor et a1,5 analyzing data from the 1987 National Health
Ruchlin
Cancer Screening in Older Women
19
Table
3. Multivariate
Correlates
of Screening
for
Cancer Woman knows
Variable
to examine
Age 65-74 Age 75-84 Age over 84 White 1 or more years of college Lives in central city of a metropolitan statistical area Participates in labor force Lives in Midwest Lives in South Lives in West Health status good Health status fair or poor Have diabetes Ever had hypertension Overweight Respondent answered all questions Believes smoking definitely increases chance of developing heart disease Believes smoking definitely increases chance of developing lung cancer Believes smoking definitely increases chance of developing emphysema Believes heavy drinking increases chance of getting cirrhosis of the liver Believes being very overweight definitely increases chance of heart disease
how breasts
Woman ever had a mammogram
Papanicolaou smear within last 3 years
0.73 0.41 0.29 1.90 1.57 1.23
(0.50,0.97) (0.16, 0.67) (0.00, 0.68) (1.67, 2.14) (1.35, 1.79) (1.01, 1.45)
0.76 0.57 0.22 1.30 1.75 1.28
(0.58,0.94) (0.39,0.75) (0.00,0.55) (1.08, 1.52) (1.61, 1.89) (1.10, 1.46)
0.64 0.46 0.33 1.00 1.57 1.21
(0.42,0.86) (0.21,0.72) (0.00,0.70) (0.69, 1.31) (1.35, 1.79)* (1.03, 1.39)
1.09 1.45 1.12 1.25 0.79 0.72 1.08 1.14 1.15 1.06 1.48
(0.86, 1.33) (1.14, 1.76) (0.89, 1.36) (0.97,1.52) (0.57, 1.01) (0.47, 0.98) (0.77, 1.39) (0.96, 1.32) (0.95, 1.35) (0.75, 1.37) (1.26, 1.70)
1.05 0.85 0.80 1.27 0.99 0.88 0.96 1.23 1.02 0.96 1.31
(0.89, 1.21) (0.65, 1.05) (0.60, 1.00) (1.07,1.47) (0.85, 1.13) (0.72, 1.04) (0.74, 1.18) (1.11, 1.35) (0.90, 1.14) (0.76, 1.16) (1.15, 1.47)
0.95 1.17 1.34 1.28 1.07 0.87 0.98 1.13 1.13 0.70 1.04
(0.72, 1.19) (0.94, 1.41) (1.11, 1.58) (1.03, 1.54) (0.87, 1.27) (0.64,l.ll) (0.67, 1.29) (0.95,1.31) (0.95, 1.31) (0.39,l.Ol) (0.82,1.26)
0.91 (0.60, 1.22)
1.12 (0.90,1.34)
1.03 (0.76, 1.30)
1.38 (1.11,1.65)
1.34 (1.12, 1.56)
1.13 (0.86,1.40)
1.13 (0.88, 1.39)
1.17 (0.97, 1.37)
0.90 (0.65, 1.16)
1.52 (1.30, 1.74)
1.08 (0.94, 1.22)
1.07 (0.85,1.29)
Data are presented as odds ratio (95% confidence interval). * Odds ratios are statistically significant at P < .05
Interview Survey and the 1984-1988 Longitudinal Study of Aging, noted that after controlling for sociodemographic factors, health status was unrelated to having a mammogram. Urban and colleagues,” reporting on the results of a 1989 telephone survey of 1538 women between the ages of 50 and 75, noted that educational attainment was not significantly related to having a mammogram within the last 2 years. Analyzing 1991 Medicare part B data for women above the age of 64 living in ten states across the country, Burns et all* noted that women in the age categories 75-84 and 84 and older reported a lower use of mammography than did those in the 6574 age group. They also reported lower use rates for nonwhite women. The important role played by physicians in the use of these screening tests is directly and indirectly noted in this study. Data reported in Table 2 under the category “Most important reason why no recent mammogram” indicate that over a third of the women noting the absence of this preventive technique claimed that it was not recommended by a physician. This finding and others reported in the literaturer3~i6 indicate that physicians who perceive themselves as rendering primary care to women need to be advocates for such important
20
Ruchlin
Cancer
Screening
in Older
Women
life-saving procedures. Medicare’s recent coverage of mammography and periodic screening for cervical cancer, should facilitate older women’s access to these important procedures, even though only a small number of respondents noted that financial reasons deterred them from having a mammogram (Table 2). If such coverage is not routinely available to women younger than 65, major efforts should be launched to remedy this deficit. Two limitations regarding the data set used in this study merit attention. First, all data in the Health Promotion and Disease Prevention Surveys are selfreported. However, for population surveillance of mammography, self-reported data are generally valid.17 Second, the data set does not permit one to recognize or control for differential mortality due to breast or cervical cancer, a phenomenon common to most existing data sets for all age groups.
References 1. US Department 2000: National
of Health and Human Services. Healthy health promotion and disease prevention
Obstetrics
people objec-
6 Gynecology
2. 3. 4.
5. 6.
7.
8.
9.
10.
11.
12.
VOL.
tives. DHHS publication no. (PHS) 91-50212. Washington, DC: US Government Printing Office, 1991. Special Medical Reports. Recommendations from AHCPR mammography guideline. Am Fam Phys 1994;50:182~5. Report of the US Preventive Services Task Force. Guide to clinical preventive services. Baltimore: Williams & Wilkins, 199673-87. American College of Obstetricians and Gynecologists. Carcinoma of the breast. ACOG technical bulletin No. 158. Washington, DC: American College of Obstetricians and Gynecologists, 1991. Mor V, Pacala JT, Rakowski W. Mammography for older women: Who uses, who benefits? J Gerontol 1992;47:43-9. Kerlikowske K, Grady D, Rubin SM, Sandrock C, Emester VL. Efficacy of screening mammography: A meta-analysis. JAMA 1995;273:149-54. Muller C, Mandelblatt J, Schechter CB. Costs and effectiveness of cervical cancer screening in elderly women. OTA-BP-H-65. Washington, DC: US Government Printing Office, 1990. Piani A, Schoenbom C. Health promotion and disease prevention: United States. 1990. DHHS Publication No. (PHS) 93-1513, Series 10, No. 185. Washington, DC: US Government Printing Office, 1993. Maddox GL. Intervention strategies to enhance well-being in later life: The status and prospect of guided change. Health Serv Res 1985;19:1007-18. Office of Technology Assessment. The use of preventive services by the elderly. Preventive health services under Medicare, paper #2. Washington, DC: US Government Printing Office, 1989. Urban N, Anderson GL, Peacock S. Mammography screening: How important is cost as a barrier to use? Am J Public Health 1994;84:50-5. Bums RB, McCarthy EP, Freund KM, Marwill SL, Shwartz M, Ash A, et al. Variability in mammography use among older women. J Am Geriatr Sot 1996;44:922-6.
90, NO.
1, JULY
1997
I3 14.
15.
16. 17.
Mandelblatt JS, Phillips RN. Cervical cancer: How often-and why-to screen older women. Geriatrics 1996;51:45-8. Breen N, Kessler L. Changes in the use of screening mammography: Evidence from the 1987 and 1990 National Health Interview Surveys. Am J Public Health 1994;84:62-7. Skinner CS, Strecher VJ, Hospers H. Physicians’ recommendations for mammography: Do tailored messages make a difference? Am J Public Health 1994;84:43-9. Kruse J, Phillips DM. Factors influencing women’s decision to undergo mammography. Obstet Gynecol 1987;70:744-7. Zapka JG, Bigelow C, Hurley T, Ford LD, Egelhofer J, Cloud WM, et al. Mammography use among sociodemogaphically diverse women: The accuracy of self-report. Am J Public Health 1996;86: 1016-21.
Address
reprint
requests
Hirsch Cornell
to:
S. Rucklin, PkD University Medical 2300 York Avenue, Box 4 New York, NY 10021
College
Received December 27, 1996. Received in revised form March Accepted March 18, 1997.
13, 1997
Copyright 0 1997 by The American College of Obstetricians Gynecologists. Published by Elsevier Science Inc.
Ruchlin
Cancer
Screening
in Older
Women
and
21