Results

Of 1334 eligible women (French speaking, alive, and Geneva residents), 958 responded (71.8%). Data from the Population Bureau indicated that eligible participants were younger on average than non-participants (mean: 56.4 years versus 58.3 years, p = .005), and more likely to be married (62.1% versus 55.6%, p = 0.03) and Swiss (76.8% versus 62.1%, p < 0.0001). After exclusion of 49 women who had a past history of breast cancer, 909 observations remained for analysis. In this group, 895 (98.5%) women answered the item on mortality reduction. Only 19.3% stated correctly that regular screening prevents “about one fourth” of deaths from breast cancer; 29.7% thought that screening prevents “about one half” of deaths, 22.3 “about three fourths”, 26.0% “didn't know”, and 2.6% thought that screening prevents no death (table 1).

The characteristics of the study sample are presented in table 2, column 1.

LACK OF OPINION ABOUT EFFICACY

In bivariate analysis, women who gave an opinion about screening efficacy and those who “did not know” differed significantly across all characteristics examined, except family history of breast cancer, citizenship, and income (table 2). Globally, women in the no opinion group were older and less educated, tended to have less favourable attitudes toward mammography screening, declined to assess their risk of developing breast cancer relative to other women, and had had fewer contacts with gynecologists.

OVERESTIMATION OF EFFICACY

Among the 639 women who believed that mammography screening reduces breast cancer mortality, six characteristics were significantly associated with overestimation of screening efficacy (table 3): a higher score of pros and a lower score of cons, willingness to pay for a mammogram, intent to undergo a mammogram in the future, a higher subjective risk of breast cancer, and Swiss nationality.

MULTIVARIATE ANALYSIS

In multivariate analysis, not having had a mammogram in the past two years nor a recent visit to a gynaecologist, having no opinion about one's risk of breast cancer, and being of ages 70 to 80 were independently associated with lack of opinion about reduction in breast cancer mortality (table 4). Only estimating one's own risk of breast cancer as above average and being Swiss independently predicted overestimation of mammography benefit.

Discussion

Despite recent controversy,16 meta-analyses summarising the best available evidence have concluded that breast cancer mortality for women aged 50 to 74 years is about 25% to 30% lower in women who undergo regular mammography screening.7 17 18 In our study, only one of five women provided an estimate of benefit in this range, while one of four was uninformed about mammography screening efficacy and one of two estimated that mammography screening reduces breast cancer mortality by 50% to 75%. This finding quantifies and extends to a general population a previous observation of overestimation of mortality reduction reported in a clinic-based sample of younger American women.19 Our results are also in line with a recent report that 55% of women representative of the US general population expected mammography screening to cut risk of death attributable to breast cancer by one half.20

Furthermore, our study showed that misconceptions about the percentage of breast cancer deaths prevented by mammography screening were consistently associated with attitudes toward mammography screening and with the intention to get a mammogram in the future. Hence inaccurate estimation of screening efficacy may affect women's overall perception of the balance between positive and negative consequences of mammography screening. Together with the finding that few women assess realistically the efficacy of mammography screening, this result indicates that most women in the general population are not in a position to provide truly informed consent before undergoing breast cancer screening, unless a more balanced information is provided to them in due course.

Multivariate analysis of risk factors for unrealistic estimates of screening efficacy supports the hypothesis that women receive inadequate information about mammography screening. Women who had had closer contacts with the preventive health care system were more likely to have an opinion about the efficacy of mammography screening, but did not provide more realistic estimates of mammography efficacy. Few socioeconomic variables were associated with inaccurate estimation of effectiveness. After adjustment, the oldest women (70 to 80 years old) were less likely to have an opinion about mortality reduction attributable to screening and Swiss citizens more likely to overestimate this reduction. However, multivariate models tend to favour proximate causes (in this case, contacts with providers of preventive care services and subjective risk of breast cancer) and disregard more distal causes, such as sociodemographic characteristics.21 Therefore, results of bivariate analysis should not be overlooked as they suggest that knowledge of the impact of screening on breast cancer mortality was deficient among the less educated and unmarried women.

Some limitations to this study should be considered. Our questionnaire assessed only the relative reduction of breast cancer mortality achievable through mammography screening; some women may have preferred mortality reduction on an absolute scale. Our response choices did not consider the possibilities that mammography screening may increase breast cancer mortality or prevent all deaths from breast cancer; the inclusion of these options may have changed the distribution of respondents among the other opinions. On average, non-respondents were older than respondents, and more likely to be unmarried and foreigners. These characteristics are often associated with a lower level of knowledge about health related issues. Furthermore, there was no organised programme of breast cancer screening in Geneva at the time of the survey. But previous research has shown that women from areas where such programmes have been in effect for several years are not necessarily better informed about the benefit of mammography screening.19 20 Finally, causality cannot be inferred from a cross sectional study.

Our results and those of previous research suggest that the widespread use of a marketing approach to information about mammography screening has been effective in imparting unwarranted optimism among women in the general population.5 19 20 22-24 Programme managers should take the current level of misinformation into account when designing information intended for screening participants.25-29 But our data also suggest that fully informed consent may reduce participation and hence the public health effectiveness of the programme. Therefore, we defend the opinion that evaluation criteria for screening programmes should be revised.30 Screening programmes should be judged not only by the level of participation achieved, but also by the proportion of women who were able to reach a truly informed decision about mammography screening, based on a balanced account of risks and benefits.