In this study, most recent 5-year RS of female BRC patients from Germany and the US and preceding trends were analyzed using data from the population-based Saarland Cancer Registry and the SEER 9/13 registries. In 2005–08, age standardized 5-year RS was 82.5% and 88.1% for all stages combined, 98.7% and 97.3% for localized BRC, 78.6% and 82.8% for BRC with local or regional spread and 23.7% and 26.7% for tumours with involvement of distant sites at diagnosis, respectively, in Germany and the US.
During the study period, most overall improvement of 5-year RS was observed for locally/regionally spread BRC, followed by localized tumours. As increases were more pronounced among German patients, the existing discrepancies between the two countries have decreased over time. Among patients with metastasized BRC, substantial improvement was seen only for patients below 50 years of age. The differences in stage-adjusted survival have entirely disappeared over time.
In Germany, 5-year RS and its improvement over time were inferior for patients aged 70+ years resulting both in an increased age gradient in Germany and prevailing differences in 5-year RS of elderly patients between the two countries.
The health care systems differ fundamentally in both countries. In Germany, health care is decentralized with private practice physicians and mostly public hospitals providing most ambulatory and impatient care. In 2007, 95.7% of the population had health insurance coverage through public or private health insurance
. In the US, health care facilities largely belong to the private sector and health insurance by private or government health insurance schemes covered 84.7% of the population in 2006/07
. Both countries have a high standard of living and spend more than 10% of their gross domestic product into health care
In the US, organized breast cancer screening was introduced (with regional differences and different types of programs) at the beginning of the 1990s
. Widespread use of mammography has been reported. The proportion of women aged 50 years or older reporting to have had a mammogram within the past two years increased between 1987 and 2005 from 27.4% to 68.4%
In Germany, implementation of organized mammography screening started only recently in 2003. The quality-assured programme offers women aged 50 to 69 years a screening mammography every two years. Prior to its introduction in 2006, only opportunistic screening was available in Saarland
. In 2002–04, 45.3% of women aged 55 years or older residing in Saarland reported to have had a mammography within the last two years (unpublished data from the population-based ESTHER cohort study in Saarland
[43, 44]). For 2007–08, a proportion of 53% of eligible women participating in the organized screening was reported
In the US, organized mammography screening started earlier and was intensely used even among younger and elderly women. Data on the usage of opportunistic mammography from Germany prior to the introduction of the organized screening are sparse. Based on the data available, it may be assumed that mammography usage and screening was much lower in Germany during the study among all age categories.
Higher usage of mammography in the US compared to Germany is well reflected in the presented incidence data. Higher proportions of earlier stages and a fourfold incidence of in situ tumours in the US correlate with the differences in the adoption and intensity of mammography screening
Diagnosis of in situ lesions of the breast is usually based on a microscopic examination and basis for a notification from a pathology laboratory. According to the registration practises of the SEER registries and the Saarland Cancer registry, we assumed the observed differences in incidence of in situ tumours to represent differences in screening activities (or prevalence of risk factors) rather than differences in case ascertainment.
The observed drop in incidence of invasive BRC in the US after the year 2000 may reflect the change in the prescription of hormone replacement therapy
[50, 51], whereas the implementation of mammography screening probably interacts with this effect in Germany. However, differences and changes in the prevalence of ‘traditional’ risk factors have also to be kept in mind. BRC mortality started to decline at the end of the 1980s in the US and approximately ten years later in Germany (data not shown).
To account for overestimation of survival of patients with screen detected cancers we performed analyses stratified by age and stage. This at least partly allowed combining patients with comparable stages with regard to begin of follow up, even if lead time effects remain present within all stages. Furthermore, beneficial effects of screening have even been demonstrated between clinically and screening detected tumours of same stage
The introduction of organized screening in Germany at the end of the study period, the higher intensity of mammography usage in the US on the one hand, and a much higher increase in 5-year RS in Germany on the other hand, support the presumption of advances in the treatment of the disease and adoption of treatment recommendations as a major cause of the remarkable gain in survival in Germany and advanced disease during the study period. However, increased early detection surely has interfered with these developments. As an observational study, it cannot fully resolve the contributions of improved treatment and screening as underlying reasons of the striking increase of 5-year RS in Germany.
Major advances in BRC treatment during the past 20 years included improved staging (e.g. more sensitive medical imaging and sentinel node dissection), propagation of breast conserving surgery and adjuvant radiotherapy for effective local treatment of early stage disease and new agents for chemotherapy and antiestrogen treatment, and targeted biologic agents. Treatment according to consensus recommendations and tailored to the patient's individual risk and a wider range of treatment options has increased the number of patients eligible for cancer-specific treatments.
For elderly patients, co-morbidity and differences in the delivery of cancer care are well documented (e.g. delay in seeking cancer care and influences of co-morbidity on selection of cancer treatments
, lower likelihood of receiving adjuvant radiotherapy or chemotherapy compared to younger patients, even after adjustment for co-morbidity and extent of disease
[54–57]) and may explain inferior 5-year RS and its less pronounced increase
The widespread use of mammography among elderly women in the US (the proportion of women aged 75+ years reporting to have had a mammogram within the last two years increased from 17.3% to 54.7% between 1987 and 2005)
 most likely exaggerated the increase in survival over time among these patients and may have partly concealed existing differences in age-specific survival in the US, as lead time effects and overdiagnosis particularly strongly affect survival estimates of a population segment with increased (age-related) mortality. Nevertheless, the observed differences in the survival of elderly patients between both countries – particularly pronounced in locally/regionally advanced disease – may point to possible limitations and hesitant delivery of care to elderly patients in Germany.
In the interpretation of the findings, a number of limitations and strengths should be considered. A major limitation is the restricted availability of clinical information. Although rather crude, the used staging scheme allowed survival estimation for three rather distinct groups of BRC patients with regard to available treatment options. Tumour stage and T classification were sufficiently complete.
The proportions of patients with missing stage were much higher in Germany than in the US. Survival of German patients with unknown stage was comparable to overall survival. Survival of US patients with unknown stage was much lower, possibly due to higher proportions of patients with clinical cancers of more advanced stage.
Further determinants beyond age and tumour stage, e.g. information about administered treatments, socio-economic status, or co-morbidity were not available. Neither was the method of first detection to consider screening effects as well as lead time, length time and overdiagnosis
The included registries operate as established registries with high levels of completeness of case ascertainment and follow-up
[14–16, 62]. So far, no other population-based cancer registry from Germany may provide trend data as presented in this article. Even if it constitutes only 1.3% of the overall population, the Saarland region is well representative for Germany and its health care system and the size of its population allowed analyses stratified by age and stage as most fundamental prognostic factors with sufficient precision. However, when interpreting point estimates of 5-year RS we also considered the corresponding trends, particularly for strata with small numbers of subjects.
The 17 registries currently included in the SEER programme cover approximately one fourth of the US population. The covered regions are regarded reasonably representative for the US, although the population is categorized as somewhat more urban, with less unemployment, higher education and higher smoking prevalence
[16, 63, 64]. One has to keep in mind, that such population differences might also have affected participation in mammography screening and access to cancer care of the included SEER populations
To avoid artefacts in the most recent survival trends due to the inclusion of the four registries which joined the SEER programme in 2000 (California regions, Kentucky, Louisiana and New Jersey), we restricted the analyses to the SEER 13 regions. Additionally performed analyses of these four regions provided almost identical survival estimates for 2005–08 (differences of overall, age- and stage-specific estimates ranging between 0.2 and 1.7% units; data not shown).
The validity of cancer diagnoses may be considered high, as almost all tumours were microscopically confirmed. When interpreting stage-specific survival, stage migration (a shift of classification towards more advanced stages due to increased sensitivity and improved accuracy of diagnostic procedures) has to be kept in mind
. Interim revisions of the BRC classification schemes
[17–19] might be of little effect only with regard to the used clinical stages.
In addition to comparability issues of included patients and tumour stages, the completeness of follow-up and estimation of background mortality is crucial in survival comparisons.
The proportions of DCO notified cases were rather small in the used databases. Under such circumstances, the effects of the exclusion of DCO cases from the analysis or a possible correction for DCO cases are expected to be very small according to recent work
RS as a measure for survival in the hypothetical situation where the disease under study would be the only cause of death (denoted as net survival) is dependent on the general population mortality (the composition of cancer patients under follow-up changes over time with regard to demographic characteristics)
. To account for such dependencies and minimize effects we provided and used age standardized estimates for inter country comparison.
The strategy of using the period analysis approach for all subsequent calendar periods aimed at deriving most up-to-date estimates (even in case of ongoing improvement) and consistent use of methodology to prevent potential artefacts in the derived time trends due to a mixture of classical cohort based and period survival estimates.
The observed trends of RS and corresponding p-values were considered on their own and carefully interpreted in the light of further empirical observations. Therefore, we did not account for multiple testing despite the overall number of performed statistical tests for significance
Follow-up was almost complete for both the German and the US patients. It has been shown, that national life tables may not adequately describe overall mortality of a subpopulation and result in biased estimates of RS (e.g. for the US, use of national life tables overestimates SEER survival estimates)
. Therefore, region- and race-specific life tables were derived for the analyses.
As the aim of population-based survival analyses is to provide a picture of the overall control of cancer in the societies compared, we did not restrict the analyses to any ethnic subpopulation (e.g. inclusion of white US patients only). Otherwise this would result in an inappropriate selective exclusion of socio-economically less privileged patients in one of the countries compared. Additional analyses were performed with a restriction to white US patients only. The effects of such an exclusion were generally small (estimates of crude, age standardized and stage adjusted survival estimates of whites were 1.5% units above estimates of all races combined), but considerable effects were observed for younger patients and more advanced stages (e.g. survival of whites aged 15–49 years with metastasized BRC was 4.5% units higher than estimates of all races combined; data not shown).