The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the National Cancer Institute/National Institutes of Health or the National Center for Health Statistics/Centers for Disease Control and Prevention.
Mammography screening allows for the early detection of breast cancer, which helps reduce mortality from breast cancer, especially in women aged 50 to 69 years. For this report, the authors updated a previous analysis of trends in mammography using newly available data from the National Health Interview Survey (NHIS).
NHIS data from 2008 were used to update trends in rates of US women who had a mammogram within the 2 years before their interview, and 2 methods of calculating rates were compared. The authors focused particularly on the 2000, 2005, and 2008 mammography rates for women aged ≥40 years, 40 to 49 years, 50 to 64 years, and ≥65 years according to selected sociodemographic and healthcare access characteristics.
For women aged 50 to 64 years and ≥65 years, the patterns were similar: Rates rose rapidly from 1987 to 2000, declined, or were stable and then declined, from 2000 to 2005, and increased from 2005 to 2008. Rates for women aged 40 to 49 years rose rapidly from 1987 to 1992 and were relatively stable through 2008. There were large increases in mammography rates among immigrants who had been in the United States for <10 years, non-Hispanic Asian women, and women aged ≥65 years who were without ambulatory care insurance.
In 2005, after more than a decade of increasing use of mammography screening in the United States, the rate of mammography use declined.1 Mammography screening detects breast cancer early and, with appropriate follow-up treatment, reduces mortality from breast cancer. The decline in rates raised the concern that mammography use, after many years of increase, was now on the decline in the United States. In this report, we examined whether the decline continued in 2008.
The mortality benefit from mammography for average-risk women in different age groups has been the focus of ongoing discussion in the mammography literature for decades.2-6 We focus largely on US Preventive Services Task Force (USPSTF) guidelines, because they are based on the current best scientific evidence. USPSTF guidelines consistently have recommended routine screening for women aged 50 to 69 years, because a mortality benefit has been observed consistently for this age group in clinical trials. For women aged 40 to 49 years and ≥70 years (or, more recently, women aged ≥75 years), the USPSTF guidelines have not recommended mammography routinely but have encouraged patient-physician discussion to determine the need and frequency for mammography.7, 8 Although the relative risk of breast cancer for women aged 40 to 49 years is about the same as that for women aged 50 to 69 years, mammograms are not as accurate in younger women, in general because younger women have denser breast tissue (more glandular, less fat). Breast tissue becomes less dense as women age, especially after menopause, which occurs on average around age 50 years.
This difference in breast density leads to differences by age in the number of breast cancers that are detected with mammography. Data pooled from clinical trials for the USPSTF analysis, which was conducted for the most recently released mammography guidelines, indicated that the number of women needed to be invited to screening to prevent 1 cancer death was 1904 for women aged 39 to 49 years, 1339 for women aged 50 to 59 years, and 377 for women aged 60 to 69 years.9 A meta-analysis by Hendrick et al indicated that there was a mortality benefit for women aged 40 to 49 years only in clinical trials that had followed them for >10 years, and the benefit was observed only after the women had reached their 50s.5 The lack of a recommendation for mass screening of older women has led to further inquiry,10 but it has not led to the controversy that has been evident regarding the screening of women aged 40 to 49 years.
MATERIALS AND METHODS
We analyzed data from the National Health Interview Survey (NHIS), a large-scale household interview survey of a statistically representative sample of the US civilian, noninstitutionalized population (http://www.cdc.gov/nchs/nhis.htm accessed October 22, 2010). The NHIS in-person interviews yield demographic and health data for all members of each participating family, and additional questions are asked about a randomly selected child (the “sample child”) and about a randomly selected adult (the “sample adult”) in each family.
In the 2000, 2003, 2005, and 2008 NHIS (and in the 2010 NHIS, which was in the field at the time this report was being written), female sample adults aged ≥30 years were asked, “Have you EVER HAD a mammogram?” Women who responded affirmatively were asked additional questions to determine the date of or time since their last mammogram. Three different formats were available for reporting that information to maximize the amount and precision of information obtained.
In the 2000 and 2003 NHIS, the questioning was as follows: “When did you have your MOST RECENT mammogram?” Respondents could answer using either Format 1 or Format 2:
In Format 1, respondents were asked to provide the DATE (month and year) of their last mammogram. If the year was provided, then this sequence of questions was ended whether or not the month was provided. If the year was not provided, then questioning went on to use Format 3.
In Format 2, respondents were asked to provide the NUMBER OF TIME UNITS since their last mammogram (in days ago, weeks ago, months ago, or years ago). If the number of time units was not provided, then questioning went on to use Format 3.
In Format 3, respondents were asked to identify the TIME INTERVAL since their last mammogram (≤1 year ago, >1 year but ≤2 years ago, >2 years but ≤3 years ago, >3 years but ≤5 years ago, or >5 years ago).
For 2005 and later, extra questioning using Format 3 was added in some cases to obtain extra information: If Format 1 was used and the month was not reported, then questioning went on to use Format 3. If Format 2 was used and the answer was reported in numbers of years ago, then questioning went on to use Format 3.
The following 2 methods were used in the current analysis to calculate rates of women having mammograms in the last 2 years.
Method I, which may be used for all data years, ignores any extra information resulting from questionnaire changes that were made starting in 2005 in the interest of using the same computational procedures across years. Also, if the year but not the month of the last mammogram was provided using Format 1, then it is imputed that the month is July.
Method II may be used for 2005 and later. If Format 3 was used, then the resulting information is used to calculate rates in the interest of using the most precise information available in each year. Breen et al1 compared mammography rates in 2000 with mammography rates in 2005 using Method I for 2000 data and Method II for 2005 data.
We defined “recent” mammograms as mammograms that were received within the last 2 years. We calculated rates of women having recent mammograms in the Years 2000, 2005, and 2008 for 4 age groups (≥40 years, 40-49 years, 50-64 years, and ≥65 years) and by selected demographic, socioeconomic, and healthcare use characteristics, which are listed in Tables 1 and 2. These same characteristics were presented in the earlier report by Breen et al1 along with the rationale for choosing them.
Table 1. The Percentage of Women Aged ≥40 Years Who Had a Mammogram Within the Last 2 Years According to Selected Characteristics From the National Health Interview Survey 2000, 2005, and 2008, Using Method I for All Years
2000 NHIS (I)
2005 NHIS (I)
2008 NHIS (I)
2000 to 2005
2005 to 2008
2000 to 2008
NHIS indicates National Health Interview Survey; CI, confidence interval; (I), Method I (see text); Unwtd No., unweighted sample size; Wtd, weighted; AA, some college or associate degree; BA, Bachelor of Arts; BS, Bachelor of Science; HMO, health maintenance organization; AIAN, American Indian/Alaska Native; Ob/Gyn, obstetrician/gynecologist.
The difference between estimates was significant at the P=.05 level.
Except for the age-group variable, estimates were standardized to the projected 2000 US population by 5-year age groups.
Table 2. The Percentage of Women Aged ≥40 Years Who Had a Mammogram Within the Last 2 Years According to Selected Characteristics From the National Health Interview Survey 2005 and 2008 Using Method I for 2000 and Method II for 2005 and 2008
NHIS indicates National Health Interview Survey; CI, confidence interval; (I) and (II), Methods I and II, respectively (see text); Unwtd No., unweighted sample size; Wtd, weighted; AA, some college or associate degree; BA, Bachelor of Arts; BS, Bachelor of Science; HMO, health maintenance organization; AIAN, American Indian/Alaska Native; Ob/Gyn, obstetrician/gynecologist.
For estimates based on the 2000 NHIS, see Table 1.
The difference between estimates was significant at the P = .05 level.
Except for the age-group variable, estimates were standardized to the projected 2000 US population by 5-year age groups.
To calculate family income as a percentage of the poverty level, family income values were imputed to replace unreported income values.11 This accounts for differences between the sample sizes for mammography rates by poverty ratio in Breen et al1 (in which the rates were calculated using only reported income values) and the sample sizes in the current analysis. Both race/ethnicity and race irrespective of ethnicity are presented. Slight differences between 2005 sample sizes in Breen et al1 using Method II and 2005 sample sizes in the current report using Method II were because 3 observations were deleted in the latter study for minor technical reasons.
Figure 1 displays age-specific, recent mammography rates in the years between 1987 and 2008 for which NHIS mammography data were available, for 4 age groups (Tables 1 and 2 provide the rates for 2000, 2005, and 2008). Confidence intervals are indicated by vertical lines through the points in Figure 1. For women aged 50 to 64 years and ≥65 years, the patterns were very similar: Rates rose rapidly from 1987 to 2000, declined, or were stable then declined, from 2000 to 2005, and increased from 2005 to 2008. The rates for women aged 40 to 49 years rose rapidly from 1987 to 1992 and were relatively stable through 2008. For all women aged ≥40 years combined, the rates declined from 2003 to 2005 and rose from 2005 to 2008. Only in the group aged 40 to 49 years did the rates decline between 2005 and 2008.
Table 1 presents age-standardized, recent mammography rates for women aged ≥40 years in 2000, 2005, and 2008 for selected demographic, socioeconomic, and healthcare access characteristics. All estimates were calculated using Method I, ie, using the same computational method for all 3 survey years. From 2005 to 2008, the overall rate did not change significantly. Significant changes over that period were experienced by only 6 of the subgroups; mammography rates rose for 3 subgroups and fell for 3 others. Declines occurred among women with family incomes that were from 200% to 299% of the federal poverty level (FPL), who experienced a 5 percentage point decline in rates, and among American Indian and Alaska Native (AIAN) women, regardless of whether race/ethnicity or race alone was used to categorize them. Rates among non-Hispanic AIAN women dropped by almost 24 percentage points, and rates among all AIAN women dropped by 18 percentage points. Rates increased by about 8 percentage points for women aged ≤64 years who had public health insurance only. The rates increased by almost 12 percentage points for Asian women; whether race/ethnicity or race alone was used to measure Asian Americans, this finding was consistent.
Table 1 also compares the mammography rates in 2000, when rates for women aged ≥50 years peaked after years of increasing (see Fig. 1), with the mammography rates in 2008. Overall mammography rates did not change significantly from 2000 to 2008. Seven subgroups experienced declines in mammography rates, and 3 subgroups had increases. Women aged 50 to 64 years had declines of about 3 percentage points. Declines of a similar size were noted for women who had family incomes ≥500% of the FPL; for women aged <65 years who had private, non-health maintenance organization (HMO) health insurance; for non-Hispanic white women; and for white women. The rates for women who were born in the United States and for women who had seen or talked to a physician in the past 12 months declined by 2 percentage points. Women in 3 smaller subgroups experienced increases in mammography use. Among women aged ≥65 years, most of whom are covered by Medicare, mammography use rose by 28 percentage points for the small group (unweighted sample sizes of 50 in 2000 and 31 in 2008) without insurance or with Medicare Part A only. Rates rose by nearly 10 percentage points for non-Hispanic Asians and by 16 percentage points among immigrants who had lived in the United States for <10 years.
Table 2 presents age-standardized rates for the same age groups and characteristics that are presented Table 1. However, the results in Table 2 were calculated using Method II for 2005 and 2008; ie, using the most information available for each of the 3 years. Table 2 replicates the comparisons of 2000 through 2005, as reported by Breen et al.1
It is noteworthy that results were consistent between the 2 tables, in the sense that, whenever a difference in 1 table was significant, the corresponding difference in the other table, whether significant or not, was in the same direction. For example, the overall mammography rates in Table 2 declined significantly by 3.7 percentage points from 2000 to 2005 and declined significantly by 3.2 percentage points from 2000 to 2008, and the respective changes in Table 1 were a significant decline of 1.8 percentage points and an insignificant decline. In addition, Table 2 reflects more significant differences than Table 1; eg, in Table 2, 17 significant decreases are listed among the subgroups from 2000 to 2005, and 13 significant decreases are listed from 2000 to 2008, whereas Table 1 lists only 9 decreases and 7 decreases for the respective periods.
Whether mammography rates had peaked and subsequently would fall was a concern when a decline in rates was discovered in the 2005 data. We assessed trends in mammography rates using recently released 2008 NHIS data. To do this, we used 2 analytic methods and compared them. We will use Method II in the future, because we prefer to use the most information available and the most precise information available. However, for the current report, we did not want our new conclusions about changes over time to be confounded by changes in the analytic methods used, so we evaluated the results from both methods. The 2008 NHIS data indicate that, overall, using either method and with exceptions for some subgroups, mammography rates did not continue to decline between 2005 and 2008. Even so, the recent 2008 mammography rates for women aged ≥40 years were below the Healthy People 2010 objective of 70%, which was met when the mammography rates peaked in 2000.
The reanalysis of trends in mammography rates attenuated the drop previously reported between 2000 and 2005, and declines previously reported for some subgroups of women became insignificant. However, both methods also indicated a significant decline in mammography use among women aged 50 to 64 years between 2000 and 2008.
Mammography use for women aged 40 to 49 years has been relatively stable since 1992 despite periodic controversy in the literature,12, 13 that made print-news headlines from time to time.13, 14 The more relaxed guidelines for women aged 40 to 49 years may help explain the stability of lower rates of mammography use in this age group.
It may be surprising to some that mammography rates declined by at least 3 percentage points between 2000 and 2008 for women aged 40 to 64 with private, non-HMO health insurance, whereas the rates increased or were stable for women with private HMO insurance. However, prevailing evidence suggests that HMOs provide better cancer screening coverage16, 17 and treatment.18
Logistic diffusion theory posits that individuals adopt new ideas or techniques in an ordered sequence. The rate of adoption over time in a social system can be displayed as an S-shaped, cumulative curve.19, 20 Historically, women with health insurance, a usual source of care, and with higher incomes and educational attainment have been early adopters of mammography.21 Between 2000 and 2008, mammography use dropped about 3 percentage points among those who had been early adopters of mammography, including women in the group aged 50 to 64 years, women with family incomes ≥500% of the FPL, women with private non-HMO insurance aged 40 to 64 years, and white women. Also surprising was a 2 percentage point drop for women who were born in the United States and for women who had had contact with a general physician in the past year. A leading hypothesis for the drop in mammography use is that some women who had visited their physician for hormone therapy (HT) prescriptions, which led to mammography referrals, stopped these visits when evidence indicated an increased risk of breast cancer associated with HT.22-24 It will be useful to have data from the 2010 cancer supplement on the NHIS to learn more about the use of mammography among these early adopters of mammography.
The few increases in mammography use between 2000 and 2008 are worth noting, because they were all ≥10 percentage points. Mammography rates rose among immigrants who had been in the United States for <10 years and among non-Hispanic Asians. There also was an increase among women aged ≥65 years who were uninsured or who had Medicare Part A only. Although these groups comprise small portions of the US population, their initial rates were low, and all of their increases were large (≥10 percentage points). Moreover, increases in these groups may be particularly important, because evidence suggests that the major reason women are diagnosed with late-stage breast cancer is that they have never been screened.25 Therefore, it is estimated that increasing rates of screening among populations with very low rates will have larger benefits than increasing the frequency of screening among women who are screened regularly.26
Public debate over the value of mammography for women aged 40 to 49 years may or may not translate into reduced rates. We observed that the rates of mammography use in women aged 40 to 49 years have remained stable since 2000 at about 64%. Rates of mammography use among women aged ≥65 years also have been stable since 2000 but at a higher level—about 67%. Whether the new mammography guidelines will affect future use remains to be determined.
In conclusion, mammography use has increased by large magnitudes in several small groups with growing populations and has declined by a small amount in large groups that historically have been early adopters. Recent changes in guidelines also may affect the rates of mammography use. If women and their physicians—and healthcare insurance companies—apply the new guidelines, then it would lead to a decline in use among women aged 40 to 49 years and an increase in use among women aged 70 to 74 years. In terms of their distribution in the US civilian, noninstitutionalized population of all women aged ≥40 years, women are divided roughly equally among the groups aged 40 to 49 years (31% in 2008), 50 to 64 years (39%), and ≥65 years (30%). This makes it difficult to predict the direction that mammography will take in the future. We look forward to the 2010 NHIS estimates, which will permit us to observe the direction that mammography use will take.
We thank Timothy McNeel of Information Management Services, Inc. for expert consultation on programming and help with updating figures and tables; Penny Randall-Levy for expert assistance with references; and Rachel Ballard-Barbash for providing a very helpful review of the article.