People in the oldest age groups are of interest in cancer prevention and control because their numbers are increasing in the general U.S. population and older people have a higher risk of dying from cancer (1). Among women, about 58.3% of deaths from breast cancer and 78.5% of deaths from colorectal cancer occur in the group aged 65 years and older (2).

Nevertheless, medical and scientific authorities continue to debate the value of routine cancer screening for women in the oldest age categories. According to screening guidelines from the United States Preventive Services Task Force (USPSTF), the evidence that supports screening mammography every 1 to 2 years for women aged 40 years and older is generalizable to women aged 70 years and older if their life expectancy is not compromised by comorbid disease (3). However, uncertainty remains about whether the potential benefits of screening mammography outweigh the harms for older women (4-7), including increased anxiety and cost from possible false-positive test results and complications from diagnostic evaluation or biopsy procedures (3).

According to USPSTF guidelines for routine colorectal cancer screening among people aged 50 years and older, the age at which colorectal cancer screening should be discontinued is not known (8). Potential complications from endoscopy (flexible sigmoidoscopy or colonoscopy) include perforation, bleeding, and infection (9).

Little is known about older women's preferred channels and methods for seeking health information and how these relate to cancer screening practices (10-14). Studies show that women aged 65 years and older are less likely than younger women to undergo cancer screening (10,13). One explanation may be that older women are unaware of the recommendations for screening. Studies show that older people are less likely than younger people to use the Internet as a source of health information (15), but we do not know what information sources older women consult most often.

If lack of knowledge about screening recommendations accounts in part for lower screening rates among subgroups of older women (16,17), learning about their preferred information sources may help to guide future interventions and improve women's health.

To learn more about cancer screening practices and the use of medical information among older women, we analyzed data from the Health Information National Trends Survey (HINTS). Our objectives were to examine 1) breast and colorectal cancer screening practices among women aged 65 years and older, and 2) the sources of health information on cancer screening practices that these women sought. The research question tested in this study was whether older women who followed breast or colorectal cancer screening recommendations preferred different sources of health information from those who did not.

About 85.0% of the women in our analysis were non-Hispanic white, 9.0% were non-Hispanic black, and 6.1% were Hispanic (Table 1). Approximately 16.9% of the women had a household income of less than $15,000 per year, and about 28.5% had less than a high school education. More than two-thirds (69.7%) had a family history of cancer. A large majority (85.9%) of the women had a usual health care provider. Almost all women (99.1%) had some type of health insurance (data not shown in table).

Data on the use of breast and colorectal cancer screening tests (Table 2) revealed that an estimated 89.4% of the women had ever had a mammogram, and 78.8% had had a recent mammogram (mammogram in the previous 2 years). Only 23.6% had had an FOBT in the previous year, and 38.8% had had endoscopy within the previous 5 years. About half (51.6%) had had an FOBT in the previous year or endoscopy in the previous 5 years.

Rates of recent mammography varied by age group, health history, and other demographic and socioeconomic factors (Table 3). Rates of recent mammography were lower for women aged 75 years and older than for those aged 65 to 74 years. Rates were much lower for Hispanic women in both age groups than for non-Hispanic women. Women in both age groups with a household income of less than $15,000 per year had low rates of recent mammography, as did those in both groups who were single (divorced/separated, widowed, or never married) and those who had 1 or no provider visits in the previous year. Older women who had no usual health care provider, especially those who were aged 75 years and older, also had low rates of recent mammography.

We also carried out a multivariate logistic analysis to see whether the associations between having a recent mammogram, income level, and having a usual health provider are confounded by having health insurance (data not shown). In a model for recent mammogram, which included age (65–74 vs 75–95 years), household income (<$15,000 vs ≥$15,000, or refused/don't know/missing), and usual provider, three variables — older age, lower income, and lack of a usual provider — were inversely associated with recent mammography (P < .03 in each instance). Health insurance coverage and education were not associated with recent mammography in this multivariate analysis (data not shown). However, almost all of the women had health insurance.

Rates of colorectal cancer screening varied by age group and other demographic factors, socioeconomic factors, and health history (Table 4). Non-Hispanic black women aged 75 years and older had the lowest rate of recent screening, and Hispanic women aged 65 to 74 years had lower rates than similarly aged non-Hispanic women. However, the numbers of women sampled in the respective categories were small. Rates of recent colorectal cancer testing were also low in each age group for women who had a household income of less than $15,000 per year, who were single, who had no family history of cancer, who had no usual health care provider, or who had 1 or no provider visits in the previous year.

In a multivariate model for colorectal cancer screening, the variables included age, education, household income, health care insurance, and a usual provider. Only lower educational attainment, lack of health insurance, and lack of a usual provider were significantly and inversely associated with colorectal cancer screening (P < .02 in each instance; data not shown). Household income was not significantly associated with colorectal cancer screening in multivariate analysis.

Data on media exposure indicate that television was the most common source of medical information for women in our analysis, regardless of whether they had recently had a mammogram or colorectal screening test (Table 5). Newspapers were the second most common source of medical information, and magazines were the third most common. Two percent of these older women did not attend to any type of media, 12% attended to one type, 16% to two types, and 70% to three or more types (data not shown in table). Women who had a recent mammogram were significantly more likely to pay attention to health information on the radio or in newspapers and magazines than were women who had not. Women who had received a recent colorectal cancer screening test were more likely to pay attention to health information in magazines or on the Internet than were women who had not.

Differences were also found between the groups' preferred channel for receiving health information (Table 5). Women who had a recent mammogram were significantly more likely to report that they wished to receive health information via personalized print, meeting with health care professionals, videocassette, audiocassette, CD-ROM, or other source than were women who had not. Women who had recently had a colorectal cancer screening test were significantly more likely to prefer receiving health information via personalized print materials or other publications (e.g., books, magazines) than were women who had not. Fewer than half of all women wanted to receive information via videocassette or audiocassette, e-mail or the Internet, or CD-ROM.

The results from this analysis of national survey data confirm and expand on results from prior studies of the use of cancer screening tests among older women (10-14,19-22), similar to the results reported by Cokkinides et al (20) using data from the Behavioral Risk Factor Surveillance System and by O'Malley et al (21) using Medicare data. The breast and colorectal cancer screening rates from the current study are slightly higher than those based on data from the National Health Interview Survey (NHIS) (22). The higher screening rates in HINTS may be explained by the use of household interviews to supplement telephone survey data in NHIS.

Culturally sensitive and appropriate educational materials about the value of screening mammography and colorectal cancer screening in reducing mortality from breast and colorectal cancer and about the potential benefits and harms of cancer screening may be helpful for older women, including low-income or minority women. Such materials should be available in both English and Spanish and should be modified to be relevant for older women from different ethnic and racial backgrounds. Educational materials should also be available for people with different levels of health literacy. Almost 29% of the women in our analysis reported having less than a high school education. Given that most health and medical information available to the public is written at a reading level higher than that of the average adult, this population may face challenges in accessing and understanding available printed health information (23-25).

Our finding that older women pay more attention to information disseminated on television and newspapers than through radio and the Internet is consistent with research suggesting that the Internet may not be the most efficient means for providing educational materials to older women (15). Our analysis also suggests that personalized print materials, other printed media, and telephone calls may be promising avenues for future efforts to disseminate information about cancer screening to older women who do not comply with screening recommendations.

Limitations of the present analysis include the cross-sectional design of the HINTS survey and the reliance on self-reported information about cancer screening and other variables. Questions for assessing cancer-related information on the HINTS questionnaire were, however, well-established and could be answered reliably and accurately by the adult population (26). New questions underwent cognitive testing (26). A further issue is that HINTS data did not allow us to distinguish between types of cancer in family histories. Other limitations include the possibility of bias due to nonresponse, given that the overall response rate in the 2003 HINTS was only 34.5%. Furthermore, the generalizability of the results to the overall U.S. population of women aged 65 years and older is unknown because of the low response rate and because noncivilian, institutionalized women and those without a household telephone were not included. Health insurance rates in the current study are higher, and cancer screening rates slightly higher, than those based on data from NHIS (22). The present study lacked information about comorbid health conditions or life expectancy, both of which may influence physician recommendations or decisions by older women about whether to undergo routine cancer screening. Finally, the results for black and Hispanic women are limited by the small number of nonwhite, non-Hispanic respondents.

Lack of knowledge about screening recommendations may partly account for lower screening rates among subgroups of older women. Our findings, therefore, that older women who adhere to breast or colorectal cancer screening recommendations and those who do not adhere to them prefer different sources of health information may help improve the health of women in this age group. Interventions designed to convey the potential benefits and risks of cancer screening to this group may benefit from knowledge of the sources of medical information used by older women and the channels of information preferred by them.