Mammography and physician breast exams after the diagnosis of breast cancer in a twin or non-twin sister

Cancer Causes and Control, 4, 251 - 260

Mammography and physician breast exams after the diagnosis of breast cancer in a twin or non-twin sister

Jean L. Richardson, Kathleen Danley, Gencie T. Mondrus, Dennis Deapen, and Thomas Mack

(Received 21 December 1992; accepted in revised form 3 March 1993)

Reports of breast-cancer-screening behavior were collected from 591 twin sisters of women with breast cancer and 182 non-twin sisters of the same women, and compared with the patterns found by national surveys. Timeline plots indicate that prior to the diagnosis of breast cancer in a sister, these women were being screened at prevailing rates. In the year after diagnosis, the annual frequency of use by them of both mammograms and physician breast exams increased by approximately 25 percent, but in subsequent years these rates dropped to a plateau no more than 10 to 15 percent higher than the baseline. The sisters over age 60 were screened with unusually low frequency, as were those with no partner in the home, those with no regular source of medical care, and especially the siblings of cases that died soon after the diagnosis. The relatively infrequent adoption of a long-term, annual, screening pattern in the face of certain knowledge of personal high risk gives cause for concern about the effectiveness of any intervention program requiring recognition of personal high risk for effectiveness. One cause for optimism is that higher screening rates prevail among those with co-twins diagnosed since 1980, suggesting that the increase in publicity and public education in recent decades has had a beneficial impact.

Key words: Breast cancer, detection, health behavior, high risk, United States.

Introduction Sisters of women with breast cancer are about two to three times more likely to develop breast cancer than women of the same age in the general population; 1-7 identical twin sisters are at even higher risk. 8 An awareness of elevated personal risk should provide an opportunity to reduce cancer mortality by providing the motivation for early detection. 9 Yet, there has been little systematic research on the influence of breast cancer diagnosis on the screening behavior of family m e m b e r s . ~0-~4

National estimates of the annual rate of physician breast examination (PBE) and of mammography in the United States show wide variation, reflecting differences in geography, calendar period, and method- ology. Results from the Access to Care Study (AtCS), is the National Health Interview Survey (NHIS) , 16-18 and the Breast Cancer Screening Consortium ( B C ) 19 a r e

shown in Figure 1. The NHIS found that 14.8 percent of women over 50 had a mammogram, and 32.5 percent had a PBE in the past year; the AtCS reported slightly

Authors are with the Department of Preventive Medicine, University of Southern California School of Medicine, Los Angeles, CA. Address correspondence to Dr Richardson, Department of Preventive Medicine, USC School of Medicine, 1420 San Pablo St, A-301, Los Angeles, CA 90033, USA. This work was supported by grants #CAO1203 from the US National Institutes of Heahh and #PBR-67992 from the American Cancer Society to the first author and grant #CA42581 from the NIH to the last author.

© 1993 Rapid Communications of Oxford Ltd Cancer Causes and Control. Vol 4. 1993 251

j. L. Richardson et al

80

70 60

'~- 50

40

~ 2o

~- ~o

NHIS

! BC-No Hx BC-Fam Hx ArCS Co-twins Sisters

Figure 1. Reported rate of PBE and mammography within the previous year. m, Mammogram; [], physical breast exam. Age > 50; White (AtCS CBE > 40).

higher proportions. The BC found that more of those with a family history had been screened with both mammography (44.5 percent) and PBEs (60.7 percent) than those without (32 percent and 55.9 percent respectively).

Other estimates of the prevalence of women with a family history who had ever had mammography range from fewer than 40 percent, 1',12 47 percent, 2° 49 percent (first-degree relatives only), ~3 52 percent, 21 and 81 percent? 2 These same investigators found that 19.6 percent of the subjects had only had one mammogram," that from 30 percent 2~ to 54 percent 22 had had one 'recently,' that 14 percent had had one annually, 13 and that the rate is slightly higher for younger than for older first- degree relatives. 23 These studies have not examined fluctuations in screening in the time period after the diagnosis in a family member, nor reasons for variation in screening rates.

In this paper, we compare the screening history of nationally ascertained twin and non-twin sisters of women with breast cancer to that of a national general population sample (NHIS). We also examine reasons for fluctuations and variations in the regularity of screening.

Materials and methods

In a US study of the etiology of cancer, participants were recruited through nationwide advertising for twins in whom one or both had confirmed cancer. The sub-sample of twins utilized here consisted of those 781 adult twin sisters of women diagnosed with breast cancer who had themselves been free of any cancer other than nonmelanoma skin cancer, here referred to as the 'co-twin.' Of the extensive questionnaires sent to these healthy co-twins in the Spring of 1988, 672 (86 percent) were returned. Of those, 591 women met our

252 Cancer Causes and Control. Vol 4. 1993

final selection criteria: non-Hispanic White women, 40 years of age and older, with at least a ninth grade education. They also were asked to provide the name and address of two non-twin sisters closest to them in age. The 591 co-twins identified 364 of their non-twin sisters. A short separate questionnaire was sent to each of these non-twin sisters, and of the 214 (59 percent) who completed and returned it, 182 (50 percent) were over 40 and with at least a ninth grade education.

The 1987 NHIS was a personal-interview house- hold-survey using a nationwide sample of the civilian population of the US. 15-17 In 1985, the sample was re- designed to include approximately 49,000 households and 135,000 persons and the response rate exceeded 94 percent. In order to provide a suitable group to compare with the sisters, we selected from the NHIS data all White women 40 years of age or older with at least a ninth grade education and no previous diagnosis of cancer; this resulted in exactly 4,000 women.

Questions to the twin-based samples were designed to be identical to the NHIS instrument. Respondents were asked their age in years, highest level of education attained, usual source of healthcare (yes/no), self-rat- ing of health status (excellent, very good, good, fair, poor), and marital status (married with a spouse in the home; married without a spouse in the home; widowed; divorced; separated; never married). Those married and living with the spouse were considered to have a partner.

After being provided with a description of PBE and mammography, subjects were asked about each pro- cedure; whether they had ever heard of it, whether they had ever used it, the length of time elapsed since the last exam, and the number of exams since the twin was diagnosed.

The responses were compared with contempor- aneous criteria for adherence to screening guidelines as follows: annual PBE for all women over age 40, mammogram annually for all women aged 50 or over, and every three years for those 40 to 49 (while the NHIS question referred only to 'within the last 3 years,' we have assumed this to be equivalent, i.e., that the use rate was distributed uniformly and at 'steady state' during that period). The age of subjects was categorized as of the time of response.

In addition to these categorical reports of screening practice, 521 (88 percent) of the twin sisters filled out a timeline describing their past screening activity, begin- ning with a point 10 years before the year of diagnosis in the twin and ending with the date of the response.

On the basis of the timelines, the proportion of sisters annually screened was calculated for each year prior to or subsequent to the date of diagnosis in the twin. The time-specific annual-screening proportions

were stratified further by the following covariates: age at diagnosis (Figure 2); decade of diagnosis (Figure 3); and survival status of the affected twin as of the date of response (Figure 4). Annual estimates based on fewer than 10 women were excluded.

We compared the time elapsed since last screening within each age stratum with the NHIS results. Because the annual rate of screening gradually fell over a period of three years following the proband twin's diagnosis, and because it is important that the NHIS results can be compared with a stabilized rate, we removed all individuals from whom responses were obtained within three years of the proband diagnosis for purposes of this analysis.

Predictors of screening behavior were examined sep- arately for the three groups: NHIS; breast cancer co- twins; and non-twin sisters of women with breast cancer. For each set, the univariate odds of having had a PBE or a mammogram within a specified time period was computed. Logistic regression was used to compare the screening rate of the co-twins or non-twin sisters with the general population, adjusting for differences between the groups with respect to indepen- dent variables: age; education; presence of a partner; and perceived health status.

Results

The demographic characteristics of the NHIS, co- twin, and non-twin sister samples are described in Table 1 (each described using information from a different source). The geographic distribution based on census categories indicates that subjects from the four major areas of the country comprise each sample in roughly equal parts. The mean ages of the groups were very similar; the NHIS includes a slightly higher proportion of women in both the youngest and the oldest categories. While the mean years of education in each group extended beyond high school, the twin sisters included a higher proportion of college graduates; they also included more married subjects and subjects with a partner in the home. Over 90 percent in each group indicated a usual source of healthcare, although NHIS respondents more often considered themselves in better health than the twin sisters. Of the latter, 54.8 percent considered themselves identical, 40.3 percent fraternal, and 4.9 percent of unknown zygosity. In two previous studies, we have found such self-designation of zygosity by adults to be very accurate. 24

The rates of PBE and mammogram use in relation to intervals prior to and after the diagnosis of cancer in a co-twin are presented in Figures 2-4. The figures separ- ately present annual screening rates from a point five

Breast cancer detection among sisters

years before the diagnosis to 15 years after the diagnosis according to age at diagnosis (40-49, 50+), decade of diagnosis (1950s-60s) and survival status.

Of the women surveyed in the NHIS, 15.9 percent reported a mammogram and 35.5 percent reported a PBE in the year prior to the survey. Of the co-twins, 10 to 15 percent had had a mammogram and 35 to 45 percent had had a PBE annually in the period prior to the diagnosis of cancer in the twin (Figure 2). Screening during the first year after diagnosis increased to a level of 35 to 45 percent for mammograms and 62 to 68 percent for PBE regardless of age. With the passage of time, the annual rate of mammography dropped to a

Table 1. Soc iodemographic characteristics of the NHIS a respondents and twins and sisters (in percents) b

NHIS Breast cancer Breast cancer (No. = 4,000) twin sister

(No. = 591) (No. = 182)

Age 40-49 30.1 19.6 15.9 50-59 22.8 25.5 24.2 60-69 23.9 37.9 32.4 70 + 23.1 16.9 27.5

Mean age 58.9 59.8 62.1 Education

Grade 9-11 17.0 9.8 NA High school

graduate 48.9 39.9 NA Some college 18.3 26.1 NA College grad 8.6 12.4 NA More than coil grad 7.1 11.7 NA

Mean grades completed 12.8 13.6 NA

Area of country Midwest 26.8 25.5 28.8 Northeast 23.2 21.0 18.1 South 29.4 26.1 26.4 West 20.5 27.4 26.9

Marital status Married 52.1 75.3 NA Separated/

divorced 16.0 8.5 NA Widowed 26.1 11.6 NA Never married 5.8 4.6 NA

Partner in the home Yes 52.1 75.3 NA No 47.9 24.7 NA

Usual source of healthcare Yes 90.3 93.6 97.8 No 9.7 6.4 2.2

Health status Excellent/very

good 56.3 35.9 28.0 Good 29.3 52.3 55.5 Fair/poor 14.4 11.7 16.5

a NHIS = National Health Interview Survey (ref. 16-18). b NA = not available.

Cancer Causes and Control. Vol4. 1993 253

J. L. Richardson et al

70 . . . . . . . . . . . . . |

so / / / ~ . - . . ~ ~ ....

~B- :2

nm 20 ' - - ~ "

1 1 ~ r I L I L I I

5 - 3 1 1 3 5 7 9 11 13 15

Years before and after twins' diagnosis

Figure 2. Association of age with annual mammography and PBE in co-twins of cases, m, PBE 40-49; +, MAM 40-49; 3~, PEE 50 +; [~, MAM 50 +. 0 = year of diagnosis, Age = age at twin diagnosis.

level of 20 to 25 percent, and that of PBE to about 50 percent for those aged 40-49, and to 35 to 40 percent for those over age 50. Thus, in the 15 years following diagnosis, annual screening rates fell about 10 to 15 percent to plateau at a level about 10 to 15 percent above the original baseline. Identical twins experienced a slightly more dramatic, immediate increase in the rate of mammography than fraternal twins; the pattern in non-twin sisters was similar to the latter.

The decade of the cancer diagnosis markedly influenced the effect on the maintenance screening rate (Figure 3). After a diagnosis in the 1950s-60s the annual mammography rate in co-twins reached only 18 percent; and after one in the 1970s it reached about 32 percent; but in the year following, a diagnosis in the 1980s reached 53 percent, a level disproportionately high in absolute terms and also in terms relative to the baseline and subsequent steady state. A similar, but less dramatic chronologic pattern can be seen in the rate of annual PBE.

The mammography screening rate was highest among those co-twins whose twin had survived to the time of response (Figure 4). Among those whose twin died within two to four years of diagnosis, the screening rate neither increased so dramatically, nor dropped so rapidly. Among those co-twins whose twin died within two years of diagnosis, the rate dropped almost to baseline just after the death of the twin.

Of the 521 co-twins who provided timeline data, 101 (19.3 percent) had a mammogram every year since the time of diagnosis, and 74 percent of these followed diagnoses made in the 1980s. While 29.5 percent of those diagnosed in the 1980s had obtained exams every year, this was true for only 9.2 percent of those diagnosed in the 1970s and 2.8 percent of those diagnosed in the 1950s-60s.

254 C a n c e r C a u s e s a n d C o n t r o l . V o l 4 . 1 9 9 3

60

50

~ 40

~. 3o

~_ 2 0

10

0 ~ r ~ i ~ p ~ L d L L d ~ ~ I ~ h I

- 5 - 2 1 4 7 10 13 Years before and after twins' diagnosis

Figure 3. Association of decade of diagnosis with rate of mammography in co-twins. D, Diagnosis 1950-69; ×, diagnosis 1970-79; ~, diagnosis 1980-89. 0 = year of diagnosis.

50

40

30 ,= o ,

~ 2 0

g_

0 i I ~ I L I i I L I ' I I I ' I ' I I

- 5 3 - 1 1 3 5 7 9 11 13 Years before and after twins' diagnosis

Figure 4. Association of case survival with rate of mammography in co-twins. D, Twin survived; ×, twin died within 2 years; ~, twin died 2 to 4 years. 0 = year of diagnosis.

Following diagnoses at ages in the 30s, 40s, 50s, and 60s, respectively, the subsequent rates at which they obtained exams every year were 12.1 percent, 18.1 percent, 23.8 percent, and 26.8 percent, respectively. A disproportionate number of the 50 co-twins reporting more than one mammogram per year were over age 50 and had a twin diagnosed in the 1980s.

Because the diagnoses occurred in different years, the timelines reflect different chronologic periods, and to ensure consistency, we compared the categorical responses with the appropriate specific years on the timeline. For example, study responses to categorical questions about activities 'last year' would roughly refer to 1987-88. The timeline and the categorical responses seem to be consistent; for example, among those who were over age 50 at the time their twin was diagnosed and whose twin was diagnosed 11 or more years ago, 15.4 percent responded to the categorical question that they had never had a mammogram, and 17.3 percent gave a comparable response using the timeline; while 9.6 percent said they had their most


Table 2. Interval since last PBE a and mammogram for NHIS b respondents and for female co-twins of women with breast cancer (percent by age)

Age c Recency PBE Mammograms

NHIS Breast NHIS Breast twin twin

40-49

50-59

60-69

70 +

Total

(No. = 1,134) (No. = 86) (No. = 1,157) (No. = 85) Never 6.4 1.2 56.8 11.8 5 + yrs 9.8 4.7 5.3 4.7 3-5 yrs 4.9 7.0 1.5 3.5 1 -< 3 yrs 36.4 16.3 18.1 21.2 < 1 yr 42.5 70.9 18.3 58.8

(No. = 869) (No. = 130) (No. = 887) (No. = 132) Never 8.6 0.8 51.1 4.6 5 + yrs 10.0 6.2 8.0 6.8 3-5 yrs 4.3 0.8 2.8 1.5 1-< 3 yrs 38.4 17.7 17.7 25.0 < 1 yr 38.7 74.6 20.4 62.1

(No. = 891 ) (No. = 205) (No. = 918) (No. = 204) Never 15.5 3.9 61.3 14.2 5 + yrs 13.1 5.4 6.5 7.8 3-5 yrs 4.6 4.9 1.6 4.4 1-< 3 yrs 33.9 15.1 15.2 17.7 < 1 yr 32.9 70.7 15.4 55.9

(No. = 863) (No. = 89) (No. = 891) (No. = 92) Never 25.5 5.6 72.3 26.1 5 + yrs 14.5 11.2 4.8 12.0 3-5 yrs 3.1 4.5 1.1 5.4 1-< 3 yrs 30.9 22.5 13.0 20.7 < 1 yr 26.0 56.2 8.8 35.9

(No. = 3,757) (No. = 510) (No. = 3,853) (No. = 513) Never 13.5 2.9 60.1 13.5 5 + yrs 11.7 6.5 6.1 7.8 3-5 yrs 4.3 4.1 1.8 3.7 1 -< 3 yrs 35.0 17.3 16.1 20.7 < 1 yr 35.5 69.2 15.9 54.4

PBE = physician breast exam. b NHIS = National Health Interview Survey (ref. 16-18). o Age refers to age at date of data collection.

recent mammogram three or more years ago, 11.5 percent reported the most recent mammogram in the cor- responding period of 1950-82; and while 42.3 percent reported a mammogram in the previous year, 34.6 percent reported having one in 1987-88. We compared the categorical and timeline data for those whose twin was diagnosed three years to 10 years prior, and for those whose twin was diagnosed three years or less prior, and found the same level of consistency.

We estimated the average regularity of postdiagnosis screening by calculating the mean annual rate in the interval after the diagnosis according to the decade in which the diagnosis occurred. Those whose twin was diagnosed from 1950-69 averaged approximately one mammogram every five years (0.17/yr); those whose

twin was diagnosed in the 1970s averaged less than one mammogram every three years (0.31/yr), and those diagnosed in the 1980s averaged one mammogram every two years (0.49/yr); the comparable annual rates for PBE were 0.48, 0.56, and 0.68, respectively. To ensure that these estimates were not influenced unduly by the more intensive screenings just after the diagnosis, more heavily represented among the co-twins of women diagnosed just before the study, we verified that the chronologic differences could be seen when the comparisons were restricted to years well after diagnosis. For example, the co-twins of women diagnosed in 1980-82 obtained mammograms in 1985 and 1986, respectively, at annual rates of 47.5 percent and 50.8 percent, and those of women diagnosed in 1970-72

Cancer Causes and Control. Vol 4. 1993 2 5 5


obtained them in 1975 and 1976 at annual rates of 13.3 percent and 23.3 percent, respectively.

The prevalence of past PBEs and mammograms for NHIS and sister respondents was compared by the age at questionnaire response, each time excluding those

sisters who completed the questionnaire within three years of the diagnosis, to estimate more accurately the comparable level at postdiagnosis steady-state.

Figure I shows a comparison of past annual-screening rates between sisters of women with breast cancer

Table 3. Odds ratio (OR) a of having a mammogram and PBE ~ in relation to demographic characteristics

NHIS ~ Breast cancer twins

OR OR OR OR Unadjusted Adjusted Unadjusted Adjusted

No mammogram within the (No. = 4,000) (No. = 542) past 3 years

Age (40-49/50 +)o 1.3 e 1.2 1.4 1.2 Source of care (yes/no) 2.0 e 2.0 e 2.2 e 2.2 Partner in home (yes/no) 1.5 e 1.4 e 1.8" 1.8" Education (any college/

grade 9-11 ) 2.3 e 2.2" 2.3 ~ 2.2" Education (any college/HS

grad) 1.4 * 1.4" 1.5" 1.6" Health status (very good,

excellent/poor, fair) 1.4 * 1.1 1.6 1.2 Health status (very good,

excellent/good) 1.1 1.0 1.3 1.2

Over age 50, no (No. = 2,794) (No. = 475) mammogram within the past year

Age (50-64/65 +)e 1.9" 1.6 e 1.4" 1.3 Source of care (yes/no) 2.9 ~ 2.8 e 2.1" 2.1, Partner in home (yes/no) 1.8 ~ 1.5 e 1.6 ~ 1.5 ~ Education (any college/

grade 9-11 ) 2.2 * 2.0" 1.3 1.2 Education (any college/HS

grad) 1.4 * 1.4" 1.0 1.1 Health status (very good,

excellent/poor, fair) 1.6 e 1.4 1.9 1.6 Health status (very good,

excellent/good) 1.0 1.0 1.1 1.0

No clinical breast exam (No. = 4,000) (No. = 542) within the past year

Age (40-49/50 +)c 1.5* 1.4 e 1.1 0.9 Source of care (yes/no) 2.7" 2.7 ~ 2.8" 2.9" Partner in home (yes/no) 1.3" 1.2" 1.6" 1.6 Education (any college/

grade 9-11 ) 1.6 ~ 1.6" 1.9" 1.7 Education (any college/HS

grad) 1.0 1.2" 1.4 1.4 Health status (very good,

excellent/poor, fair) 1.2" 1.0 2.3" 2.0 e Health status (very good,

excellent/good) 1.0 0.9 1.6" 1.6 ~

a To compute the odds ratios, each variable is categorized as study group/reference group as noted in parentheses for each variable.

b PBE = Physician breast exam. o Age refers to age at date of data collection.

NHIS = National Health Interview Survey (ref. 16-18). e p~< 0.05.


and subjects in national samples. The rates found in this study are comparable to the rates for those with a family history in the BC study. Because both the BC and AtCS studies used a telephone survey, these rates might be inflated artificially over the NHIS rates. Regardless of age, both twin sisters and non-twin sisters of women with breast cancer were more likely to have ever had a PBE and a mammogram, and more likely to have had one within the past year, than NHIS respondents (Table 2). There was no significant differ- ence between identical and fraternal twins. More than 95 percent and 84 percent, respectively, of each group of sisters had had a PBE or a mammogram at some time in their lives, whereas less than 87 percent and 40 percent of the general population had done so. However, despite the evident familial risk, 45.6 percent of breast cancer twins and 48.3 percent of non-twin sisters had not had a mammogram, and 30.8 percent of the twins and 39.7 percent of the non-twin sisters had not had a PBE within the past year (the comparable figures from the general sample are 84.1 percent and 64.5 percent). Regardless of family history, older women were less likely ever to have had either a mammogram or clinical breast exam.

We compared the observed screening behavior with that recommended in relation to age, education, a regular source of healthcare, a partner in the home, and health status. Results from the twin sisters were the same as those from the NHIS respondents; these univariate analyses are presented in Table 3. Younger women were more likely to have received the recommended screening by PBE and mammography, as evidenced among the sisters and more strongly among the NHIS respondents. Subjects with a usual source of healthcare were more than twice as likely to have had timely breast exams and mammograms, and those with a partner in the home were also significantly more likely to have done so. Women who did not finish high school were somewhat less likely to have been screened, but those who had some college education obtained screening no more often than other high school graduates. Perceived health status was not a very accurate predictor of regular screening although the association was in the appropriate direction.

We then computed the odds (after adjustment for age, healthcare, having a partner, education, and health status) of being screened after the diagnosis in a twin, using the NHIS sample as a referent standard. This gives us an estimate of the effect of the twins' breast cancer diagnosis on the screening behavior of the un- affected twins. Four adjusted odds ratios were computed; the adjusted odds of having a mammogram within the past three years for twins of women with breast cancer was 5.9 times higher for those aged 40-49,


and 6.2 times higher for those over age 50. For those aged 50 and over, the adjusted odds of having a mammogram within the past year were 5.8 times higher and, for all ages combined, the adjusted odds of a co-twin having a clinical breast exam within the past year was 4.0 times higher than the expectation. Each estimate of relative odds calculated after having substituted the non-twin sisters for the co-twins was nearly the same.

Discussion

Twins, especially identical twins, are usually dose and quite aware of their similarity. Whether or not breast cancer is thought to be inherited, the appearance of breast cancer in one twin carries a message of high risk to the other twin in a most direct and convincing man- ner, and the message is at least equivalent to an effective educational message about personal risk. If cognitive appreciation of this risk provides an important incen- tive for regular screening, then a change in such behavior should follow the diagnosis of breast cancer in a twin. 25

Our findings suggest that this is superficially true; by the time of response, nearly 87 percent of these co- twins and 84 percent of the non-twin sisters had had at least one mammogram in comparison with 40 percent of the general population. These rates are far higher than previously reported rates of 35 percent to 52 percent of women with a family history (usually a sister) of breast cancer. 11-12,2°-22 The clear surge in screening behavior after the proband's diagnosis strongly suggests that recognition of self-risk played an important role.

However, if it is true that those confronted with their own potential morbidity are more likely to promptly adopt a screening behavior, it is also true that after the initial surge of screening behavior, even these volunteers failed to comply with the recommended frequency of both PBEs and mammograms, and the level of compliance continued to decrease with the passage of time. Even though incidence is linked directly to increasing age, and the screening habits of older co-twins of women with breast cancer were far better than those of older women in general, the sisters who were aged 60 and over obtained fewer mammograms and PBEs than those who were under age 60.

It is true that those made aware of their high risk in the 1980s were screened more frequently in both the short- and long-term than those in the 1970s and cer- tainly more frequently than those in the 1950s and 60s, whether because of the increasing public profile of screening or because of an increasing awareness of the implications of cancer among family members. Even in the 1980s, however, follow-up screening activities

Cancer Causes and Control Vol 4. 1993 2 5 7


were below the level required to greatly reduce the frequency of interim diagnoses.

Screening regularity has been examined in three large mammography-efficacy trials. In the Health Insurance Plan (HIP) trial, of the women invited to attend, 67 percent appeared for the initial screen; of these, 80 percent, 75 percent, and 69 percent appeared for the sequential mammograms. 26 In the recent Swed- ish trial, attendance was 74 percent in the first round and 70 percent in subsequent rounds? 7 In the Nether- lands, compliance with the recommended four to six exams over a 10-year period declined from 87 percent to 67 percent for those under age 50 and from 83 percent to 50 percent for those aged 50 to 64. 28 Attendance rates did improve between the 1960s and the 1980s, but it is clear nevertheless that maintenance of the compliance with screening is difficult even under trial cir- cumstances, with the attendant free services and multiple follow-up reminders.

For those over age 50 at the time of the twin's diagnosis, the level of annual compliance dropped from 43 percent in the year immediately following the twin's diagnosis, to 32 percent in the fourth year, and gradually to under 20 percent, nearly the level in the general population. Only a fifth of all twins formed the habit of having an annual mammogram every year following their twins' diagnosis and only a fraction of that proportion did so before the 1980s. While even among women at such obvious high risk, strict adherence to the recommended annual guideline was uncommon, a substantial number of those over age 50 did report having had a mammogram every other year.

The secular increase in adherence amounted to about seven percent over the prediagnosis screening rate with each decade, but the figure is much higher for the period immediately following the proband diagnosis, suggesting that mammography is one immediate response to new-found awareness of personal risk. While the co-twins of women diagnosed in the 1980s reported mammograms on average every two years, the majority of this experience represents the immediate postdiagnosis period, and the observed gradual weakening of compliance with time suggests that this impressive rate will not last.

In addition to age, education, and decade, other factors associated with less frequent screening in the face of a diagnosis in the family have been found by others: 29 the lack of a regular source of healthcare; the lack of a partner in the home; and poorer health status. The nearly 25 percent differential between the frequency of mammograms and that of PBEs suggests that women who present themselves for a breast exam often are not scheduled for a mammogram. An important opportunity for mammography therefore is being lost,

2 5 8 Cancer Causes and Control. Vol 4. 1993

whether by inaction of the physician, the woman, or both.

Most surprising was the finding of a link between the survival of the affected twin and the regularity of screening. The basis of this link is unclear. Perhaps the death of the affected twin just represents the loss of a daily reminder, perhaps it marks a paralyzing increase in the level of fear, or perhaps the opposite, an increased opportunity for denial, the affected twin being out of sight and out of mind. Or finally, perhaps the loss reinforces the view that early treatment and therefore regular screening is ineffective. Of course, the truth may include all of the above.

Women who deny or ignore health risks are less likely to volunteer as these women have, and reports of compliance from volunteers may be overstated. More- over, our results from well-educated women cannot be extended to socioeconomically disadvantaged women, since a special disincentive to screening is posed by the inaccessibility and cost of preventive procedures for some under the current system of medical care in the US. Notwithstanding these limitations, these findings provide a measure of the effect of self-perception of risk.

It customarily is presumed that to preferentially serve that subgroup of women with more than their share of breast cancer risk, such women must be identified, made aware of their special status, assured of access to a screening program, and induced to take advantage of it. The relatively low proportion of high- risk women who actually have benefited from screening efforts has been well documented. 11-13 Each of these required steps has been shown to contribute to the failure of such intervention. Few physicians take the trouble to identify women at high risk, 2° too few routinely obtain the pertinent information and inform patients of their general risk level, 23 and too many of the women who are so informed have been prevented from acting by economic or logistical obstacles. H

However, even special efforts to identify such women, counsel them, and facilitate their access to screening have met with no more than modest suc- cess, 2~ indicating that the mere provision of information and services is not enough to ensure that high risk persons will take action.

Our findings give similar cause for concern, not because subjects who clearly had been made aware of their level of risk failed to obtain screening, but because they failed to maintain a program of regular screening, and because it was precisely those women who stood to gain the most from mammography, namely older women and women with a family history of aggressive disease, who were least successful. The earlier treatment of prevalent cases identified by a single screening


is not likely to provide much reduction in mortality, even among women at high risk. 27 If intervention is to become truly effective, a sustained program of repeated screening designed to identify rapidly growing lesions at an early stage will be required.

When screenees who have been informed of positive findings are followed up, the anxiety resulting from the bad news appears to interfere with compliance with medical recommendations, 2~ yet those women who cope with such anxiety by the method of denial also have been shown to seek screening less often. This has led to the suggestion that an optimal balance between fear and denial may exist, in the absence of which various self-protective behaviors more frequently are suppressed." Such a phenomenon might provide some explanation for our findings, including the observation that twins of patients with fatal breast cancer were less likely to enter into a sustained screening program. At the very least, the evidence suggests that there are multiple complex endogenous reasons for failing to respond to screening advice and this bodes ill for programs which rely solely on the well informed woman to protect herself.

We suggest that effective lifelong secondary prevention of breast cancer, like effective primary prevention, will require more than just the motivation of the sub- ject and her regular physician. It will require some for- mal means of follow-up designed to both recognize the markers of high risk and deliver regular periodic reminders to both the woman and her physician. It will be the rare primary practitioner who has the patient volume to put such a system in place, but it should not be beyond the means of highly specialized prac- titioners or groups.

References 1. Anderson D. A genetic study of human breast cancer.

JNCI 1972; 48: 1029-34. 2. Anderson D. Genetic study of breast cancer: Identifi-

cation of a high risk group. Cancer 1974; 34: 1090-7. 3. Henderson BE, Powell D, Rosario I, et al. An epidemi-

ologic study of breast cancer.JNCI 1974; 53: 609-14. 4. Albert S, Child M. Familial cancer in the general popu-

lation. Cancer 1977; 40: 1674-9. 5. Bain C, Speizer FE, Rosner B, Belanger C, Hennekens

CH. Family history of breast cancer as a risk indicator of the disease. ArnJ Epidemio11980; 111: 301-8.

6. Macklin MD. Comparison of number of breast cancer deaths observed in relatives of breast cancer patients and the number expected on the basis of mortality rates. JNCI 1959; 22: 927-51.

7. Ottman R, Pike MC, King MC, Cassagrande JT, Hen- derson BE. Familial breast cancer in a population based series. ArnJ Epidernio11986; 123: 15-20.

8. Holm NV, Hauge M, Jensen M. Studies of cancer aeti- ology in a complete population: breast cancer, colorectal cancer and leukemia. Cancer Surv 1982; 1: 17-32.

9. Schottenfeld D. Cancer detection programs. In: Frau- meni JF Jr, ed. Persons at High Risk of Cancer--An Approach to Cancer Etiology and Control. New York: Academic Press Inc, 1975: 399-413.

10. Anderson DE, Romsdahl MM. Family history: a cri- terion for selective screening. In: Mulvihill JJ, Miller RW, Fraumeni JF Jr, eds, Genetics of Hurnan Cancer. New York: Raven Press, 1977: 257-62.

11. Vogel VG, Graves DS, Vernon SW, Lord JA, Winn RJ, Peters GN. Mammographic screening of women with increased risk of breast cancer. Cancer 1990; 66:1613-20.

12. Krischer JP, Cook B, Weiner RS. Identification and screening of women at high risk of breast cancer. Cancer Detect Prev 1988; 13: 65-74.

13. Kaplan KM, Weinberg GV, Small A, HerndonJL. Breast cancer screening among relatives of women with breast cancer. Am J Public Health 1991; 81: 1174-9.

14. Dworsky R, Paganini-Hill A, Spicer D. Breast cancer screening and education in high risk women. Adv Cancer Control 1989; 115-24.

15. Hayward RA, Shapiro MF, Freeman HE, Corey CR. Who gets screened for cervical and breast cancer? Results from a new national survey. Arch Intern Med 1988; 148: 1177-81.

16. Provisional estimates from the national health interview survey supplement on cancer control. MMWR 1988; 37: 417-20.

17. National Center for Health Statistics. Current Estimates From the National Health Interview Survey, 1982: Vital and Health Statistics. Washington DC: Public Health Service, 1985; DHHS Pub. (PHS) 85-1578, Series 10, No. 150.

18. National Center for Health Statistics. The National Health Interview Survey Design, 1973-1983: Vital and Health Statistics. Washington DC: Public Health Ser- vice, 1985; DHHS Pub. (PHS) 85-1320, Series 1, No. 18.

19. NCI Breast Cancer Screening Consortium. Screening mammography: a missed clinical opportunity? Results of the NCI breast cancer screening consortium and national health interview survey studies. JAMA 1990; 264: 54-8.

20. BondyML, VogelVG, Halabi S, Lustbader ED. Identifi- cation of women at increased risk for breast cancer in a population-based screening program. Cancer Epid Bio- mark Prey 1992; 1: 143-7.

21. Houts PS, Wojtkowiak SL, Simmonds MA, Weinberg GB, Heitjan DF. Using a state cancer registry to increase screening behaviors of sisters and daughters of breast cancer patients. Arn J Public Health 1991; 81: 386-8.

22. Zapka JG, Hosmer D, Costanza ME, Harris DR, Stod- dard A. Changes in mammography use: economic, need and service factors. Am J Public Health 1992; 82: 1345-51.

23. Stefanek ME, Wilcox P. First degree relatives of breast cancer patients: screening practices and provision of risk information. Cancer Detect Prey 1991; 15: 379-84.

24. Deapen D, Escalante A, Weinrib L, et al. A revised estimate of twin concordance in systemic lupus erythemato- sus. Arthritis Rheum 1992; 35:311-8.

Cancer Causes and Control. Vo14. 1993 259

j. L. Richardson et al

25. Lerman C, Rimer BK, Engstrom PF. Cancer risk notifi- cation: psychosocial and ethical implications. J Clin Onco11991; 9: 1275-82.

26. Shapiro S, Venet W, Straz P, Venet L, Roeser R. Selec- tion, follow-up and analysis in the Health Insurance Plan Study: a randomized trial with breast cancer screening. NCI Monogr 1985; 67: 65-74.

27. Andersson I, Aspegren K, Janzon L, et al. Mammo- graphic screening and mortality from breast cancer: the

Malm6 mammographic screening trial. Br MedJ 1988; 297: 943-8.

28. Peeters PHM, Verbeek ALM, Hendriks JHCL, van Bon MJH. Screening for breast cancer in Nijmegen. Report of 6 screening rounds. 1975-1986. Int J Cancer 1989; 43: 226-30.

29. Vernon SW, Laville EA, Jackson GL. Participation in breast screening programs: a review. Soc Sci Med 1990; 30: 1107-18.


Documents

Mammography and physician breast exams after the diagnosis of breast cancer in a twin or non-twin sister