Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Treatment Strategies and Survival of Older Breast Cancer Patients – An International Comparison between the Netherlands and Ireland

  • Mandy Kiderlen,

    Affiliations Dept. of Surgery, Leiden University Medical Center, Leiden, The Netherlands, Dept. of Gerontology & Geriatrics, Leiden University Medical Center, Leiden, The Netherlands

  • Paul M. Walsh,

    Affiliation National Cancer Registry of Ireland, Cork, Ireland

  • Esther Bastiaannet,

    Affiliations Dept. of Surgery, Leiden University Medical Center, Leiden, The Netherlands, Dept. of Gerontology & Geriatrics, Leiden University Medical Center, Leiden, The Netherlands

  • Maria B. Kelly,

    Affiliation National Cancer Registry of Ireland, Cork, Ireland

  • Riccardo A. Audisio,

    Affiliations Department of Surgery, St Helens Teaching Hospital, St Helens, United Kingdom, University of Liverpool, Liverpool, United Kingdom

  • Petra G. Boelens,

    Affiliation Dept. of Surgery, Leiden University Medical Center, Leiden, The Netherlands

  • Chris Brown,

    Affiliation National Cancer Registry of Ireland, Cork, Ireland

  • Olaf M. Dekkers,

    Affiliation Dept. of Epidemiology, Leiden University Medical Center, Leiden, The Netherlands

  • Anton J. M. de Craen,

    Affiliation Dept. of Gerontology & Geriatrics, Leiden University Medical Center, Leiden, The Netherlands

  • Cornelis J. H. van de Velde,

    Affiliation Dept. of Surgery, Leiden University Medical Center, Leiden, The Netherlands

  • Gerrit-Jan Liefers

    Affiliation Dept. of Surgery, Leiden University Medical Center, Leiden, The Netherlands



Forty percent of breast cancers occur among older patients. Unfortunately, there is a lack of evidence for treatment guidelines for older breast cancer patients. The aim of this study is to compare treatment strategy and relative survival for operable breast cancer in the elderly between The Netherlands and Ireland.

Material and Methods

From the Dutch and Irish national cancer registries, women aged ≥65 years with non-metastatic breast cancer were included (2001-2009). Proportions of patients receiving guideline-adherent locoregional treatment, endocrine therapy, and chemotherapy were calculated and compared between the countries by stage. Secondly, 5-year relative survival was calculated by stage and compared between countries.


Overall, 41,055 patients from The Netherlands and 5,826 patients from Ireland were included. Overall, more patients received guideline-adherent locoregional treatment in The Netherlands, overall (80% vs. 68%, adjusted p<0.001), stage I (83% vs. 65%, p<0.001), stage II (80% vs. 74%, p<0.001) and stage III (74% vs. 57%, P<0.001) disease. On the other hand, more systemic treatment was provided in Ireland, where endocrine therapy was prescribed to 92% of hormone receptor-positive patients, compared to 59% in The Netherlands. In The Netherlands, only 6% received chemotherapy, as compared 24% in Ireland. But relative survival was poorer in Ireland (5 years relative survival 89% vs. 83%), especially in stage II (87% vs. 85%) and stage III (61% vs. 58%) patients.


Treatment for older breast cancer patients differed significantly on all treatment modalities between The Netherlands and Ireland. More locoregional treatment was provided in The Netherlands, and more systemic therapy was provided in Ireland. Relative survival for Irish patients was worse than for their Dutch counterparts. This finding should be a strong recommendation to study breast cancer treatment and survival internationally, with the ultimate goal to equalize the survival rates for breast cancer patients across Europe.


Currently, about 40 per cent of all new breast cancer cases in developed countries occur among women aged 65 and older [1]. Life expectancy is increasing, diagnostic tools become more sensitive and screening programs are more widely used and expanded. Consequently, the proportion of elderly breast cancer patients is expected to increase in the near future [2].

Proper treatment for older breast cancer patients is difficult to define. Older women are frequently excluded from clinical treatment trials because of their age, comorbidity or logistical barriers [3]. Moreover, the elderly who are included in trials are probably not representative for the general older population [4]. Consequently, an evidence-based treatment strategy for older women with breast cancer is lacking. The only guidance for clinicians is from treatment guidelines which have been validated in younger and healthier women [5]. Extrapolation from trials might not be valid since breast cancer biology differs in some respects in older patients, treatment tolerance varies, and there are substantial competing risks of mortality [2,6]. Consequently, clinicians have to decide what is best for their patient: treatment according to the guidelines, or patient-tailored deviation from the guidelines.

In the last decade it has become more accepted to use observational data, preferably population-based, to assess treatment effects in older cancer patients [7]. However, no strong conclusions can be drawn from these studies as bias due to confounding by indication is likely to be present, since specific (unknown) patient and tumor-related factors influence receipt of particular treatments.[8]

A recent observational study comparing locoregional treatment between six European countries and the US found that treatment strategy in The Netherlands and Ireland differed considerably on various items among older women with early stage breast cancer, indicating that older patients with early stage breast cancer in Ireland seemed to be slightly undertreated, compared with The Netherlands. However, relative survival was not demonstrably different [9].

The aim of the present study is to compare treatment strategy and relative survival for operable (non-metastatic) breast cancer in the elderly between The Netherlands and Ireland in more detail.

Materials and Methods


From the Netherlands and Irish cancer registry, all female patients aged 65 years and older diagnosed between 2001 and 2009 with invasive, non-metastasized breast cancer were selected. Patients with a diagnosis of breast cancer on death certificate or at autopsy only, and other patients with a survival time of zero days, were excluded. If a patient had a second primary tumor during follow-up, only the first primary breast tumor was considered for analyses.

Tumor stage was defined by TNM stage [10], with clinical T and N used when pathological information was lacking. Patients with missing T category were excluded. When nodal and distant metastatic status were unspecified (NX and MX), status was assumed to be N0 and M0, respectively. Stage data were originally coded using 6th-edition TNM [10] rules in the Netherlands and 5th-edition TNM [11] rules in Ireland. Micrometastases (≤0.2 cm) in regional nodes, classified as N1a in 5th-edition TNM were recoded to N0 for 21 Irish cases to conform to 6th-edition TNM rules. For surgical treatment, only the most extensive surgery registered was used for analysis. Axillary surgery was coded as yes or no.

Primary outcome was treatment strategy by stage. Treatments of interest were type of surgery (none, BCS or mastectomy), radiotherapy (RT; yes or no), axillary surgery (yes or no), locoregional guideline adherence (details below), endocrine therapy (yes or no) and chemotherapy (yes or no). Secondary outcome measure was 5-year relative survival in each country.

In both the Dutch and Irish breast cancer guidelines, primary surgical treatment with mastectomy or BCS followed by radiotherapy (RT) is recommended for non-metastasized breast cancer. In addition, it is recommended to assess axillary nodal status by performing a sentinel node procedure or axillary lymph node dissection [1214] (S1 Appendix). Therefore, locoregional treatment was considered guideline-adherent when a patient had BCS and RT or mastectomy with or without RT, in all cases followed by any axillary surgical procedure. In addition the receipt of systemic therapy (adjuvant endocrine therapy and chemotherapy) was analyzed.

Routine cancer registry data on endocrine therapy in Ireland were known to be incomplete (National Cancer Registry of Ireland, unpublished data), because of difficulties associated with outpatient prescription of the drugs involved. Endocrine therapy data for Irish patients were therefore supplemented by linkage to a national database of drug prescription, which covers publicly funded ‘medical card’ patients including most patients aged 65 years and over. Additional endocrine therapy was identified by this linkage for 21% of patients. Linkage was not possible for about 15% of Irish patients, and for this group, ‘missing’ endocrine therapy was imputed (4% of all patients). The imputation assumed that the proportion of ‘linked’ patients receiving endocrine therapy by stage (I, II and III), hormone receptor status (any positive vs. none positive) and broad age-group (65–74 and 75+) also applied to unlinked patients, and these ‘extra’ treatments were assigned randomly within each stage-by-age group.

Data from both the Netherlands Cancer Registry and the National Cancer Registry of Ireland are fully anonymized prior to being made available to researchers, so data cannot be traced back to the individual patient. Therefore, no informed consent was required from the included patients and there was no need for approval of an ethical committee.

Mortality follow-up was available to December 31st 2011 by linkage of cancer registry with national mortality data.

Statistical analyses

Analyses were performed in IBM SPSS Statistics 20 and Stata SE 12. Treatment strategies were analyzed grouped by tumor stage (I to III). Differences in treatment between countries were tested by a Poisson regression model, adjusted for age (continuous), histological subtype, tumor grade, ER and PR status.

Relative survival was calculated by the Ederer II method [15] as the ratio of the survival observed among the cancer patients to the expected survival based on the corresponding general population (by age, sex, and year of diagnosis), using the ‘strs’ command in Stata. National life tables for each country were used to estimate expected survival. Results were presented as percentage relative survival after five years, and Relative Excess Risks (RER) derived from relative survival modeling, with The Netherlands as reference category [16].


Overall, 41,055 patients from The Netherlands and 5,826 patients from Ireland were included. Patient and tumor characteristics are shown in Table 1. Median age for patients in The Netherlands was 74 years (range 65–102), and in Ireland 74.2 (range 65–99). Fewer early stage tumors, and more with advanced stage were observed in Ireland (P<0.001). Recorded grade distribution differed significantly, with a higher proportion of higher grades in Ireland than in The Netherlands (P<0.001).

Hormone receptor status showed smaller differences, with slightly smaller proportions of estrogen and progesterone receptor positive tumors among Irish patients (81% and 64%, respectively, excluding missing or unknown values) compared with those from the Netherlands (86% and 67%) (P <0.001). The proportion of missing values was much lower in Ireland, mainly because Dutch data were not complete for the years 2001–2005 rather than to differences in proportions of patients tested.

Locoregional treatment

Fig. 1A shows the proportions of patients receiving guideline-adherent locoregional treatment by country, grouped by stage. In The Netherlands guideline-adherent treatment was performed in 80%, with little variation between stages, whereas these proportions in Ireland ranged from 57% (stage III) to 74% (stage II). Among patients who did not receive guideline-adherent locoregional treatment, 65% (The Netherlands) and 68% (Ireland), had no locoregional treatment at all, 6% (The Netherlands) and 13% (Ireland) had only BCS (without RT or axillary surgery), and 29% (The Netherlands) and 20% (Ireland) had adequate local treatment, but no axillary surgery. Adjusted RRs for having guideline-adherent locoregional therapy in Ireland relative to The Netherlands were 0.79 (95% CI 0.76–0.81), 0.87 (0.85–0.89) and 0.72 (0.68–0.75) respectively for stage I, II and III (P<0.001 for all stages).

Fig 1. Treatment strategies.

A. Locoregional guideline-adherence by stage. B. Endocrine therapy for estrogen receptor positive patients by stage. C. Chemotherapy by stage.

Looking more specifically at locoregional treatment (Table 2), overall, more patients in Ireland had no breast surgery at all (19% vs. 12% in The Netherlands), also stratified by stage (P<0.001 in all stages).

In The Netherlands, 82% underwent any axillary surgical procedure, as compared to 74% in Ireland. Also, in all three stage groups, fewer patients in The Netherlands than in Ireland did not undergo axillary surgery (P<0.001).

Regarding radiotherapy (RT), among all patients, more patients received RT in Ireland than in The Netherlands, overall and after mastectomy (P<0.001). For mastectomy patients, the difference was only seen in stage I (18% of patients had post-mastectomy RT in Ireland vs. 3% in The Netherlands) and stage II (42% vs. 14%) (P<0.001). In stage III patients, the difference in the receipt of RT attenuated and no difference was observed in post-mastectomy RT. However, in all stages significantly fewer patients in Ireland received RT after BCS (79% vs. 94% in The Netherlands, overall, P<0.001). (Table 2).

Endocrine therapy

The overall proportion of estrogen receptor positive patients receiving endocrine therapy differed between the countries—59% in The Netherlands vs. 92% in Ireland (P<0.001) for all stages combined. Patients with stage I disease were more than three times as likely to get endocrine therapy in Ireland (91% vs. 27%; P<0.001). The difference was smaller in stage II patients, 94% in Ireland vs. 84% in The Netherlands (P<0.001), and 89% vs. 91% respectively in stage III patients (P = 0.188) (Fig. 1B; Table 2). Adjusted RRs for having endocrine therapy in Ireland were 2.91 (95% CI 2.77–3.05), 1.11 (1.09–1.12) and 0.99 (0.96–1.02) respectively for stage I, II and III ER-positive patients. Among patients who did not receive any locoregional treatment at all, the proportions of endocrine monotherapy were 85% in The Netherlands and 86% in Ireland.


Overall, 6% of patients The Netherlands and 24% of patients in Ireland received chemotherapy, and there was a higher proportion of Irish patients that received chemotherapy in all three stages (P<0.001) (Fig. 1C; Table 2). Adjusted RRs for having chemotherapy in Ireland were 4.55 (95% CI 3.81–5.43), 3.35 (3.11–3.62) and 1.44 (1.31–1.58), respectively for patients with stage I, II and III.

Relative survival

Median follow-up time was 4.5 years for The Netherlands and 4.3 years for Ireland. During the total follow-up period, 14,771 (36.0%) patients died in The Netherlands, compared to 2,191 patients (37.6%) in Ireland.

Five-year relative survival was 88.8% in The Netherlands and 82.9% in Ireland, for all stages combined (Fig. 2). This survival difference was statistically significant, also after adjustment for age, grade, stage, ER, PR and morphology (relative excess risk [RER] for Ireland, with The Netherlands as reference category: 1.22; 95% confidence interval (CI) 1.10–1.36). Grouped by stage, no survival difference was demonstrated in stage I patients (adjusted RER 1.00, 95% CI 0.59–1.70), but worse survival was confirmed for Irish patients in stage II (adjusted RER 1.20, 95% CI 1.02–1.42) and stage III (1.20, 95% CI 1.04–1.39).


The current study, comparing treatment and relative survival of older breast cancer patients between two Western European countries with similar treatment guidelines, showed large differences in treatment approach for older breast cancer patients. A higher proportion of patients in The Netherlands received guideline-adherent locoregional treatment than in Ireland in all stages, but in Ireland the receipt of systemic treatments was higher in all stages of disease. Relative survival of patients in Ireland was significantly poorer than in The Netherlands, but adjusted models suggested the difference most marked for stage II and III patients.

The observed discrepancies in breast cancer treatment are consistent with findings of earlier international comparisons of older and other breast cancer patients across Europe and the US [9,1719]. However, although international survival and treatment variations among breast cancer patients have recently been assessed on a global scale [20], no clear evidence was published on the potential role of different treatment strategies in influencing survival discrepancies among early-stage cases. However, it is interesting to speculate on reasons for the differences in patterns of care for the specific treatment modalities. Some differences could be explained by discrepancies in guideline recommendations between the two countries. There are differences (S1 Appendix) especially for postmastectomy radiotherapy and chemotherapy and for both of these, indications are broader in the Irish guidelines. These differences are reflected in our results, where we observed more patients from Ireland receiving these treatments, as compared to the patients from Netherlands, also stratified by stage of disease. Secondly, physicians from the Netherlands may also be more likely to deviate from the guidelines when treating older breast cancer patients. Unfortunately, in our study it was also not possible to draw any reliable conclusion about the impact of differences in any of the specific treatment modalities, because of a potential bias due to confounding by indication when comparing the outcomes of patients with different treatments directly.

In a large population-based study in The Netherlands, guideline adherence of breast cancer treatment among younger and older breast cancer patients was compared between different regions, and although differences in adherence were observed, there were no significant survival differences between regions [21]. In the current study we found less guideline-adherence on locoregional treatment in Ireland, and this was accompanied by a worse survival in Ireland. On the other hand, patients in Ireland received more systemic therapies (both endocrine therapy and chemotherapy), so no conclusion can be drawn based on the locoregional treatment only, because of a probably counterbalanced effect by adjuvant treatments.

To obtain the highest level of evidence on treatment benefits, the effect of each treatment modality should be investigated based on randomized assignment of treatment. However, randomized clinical trials (RCTs) tend to be slow, expensive, and insensitive to the heterogeneous contexts of the general population [22]. The disadvantages of RCTs are probably even stronger in the older population, because of their limited mobility and large heterogeneity. Observational studies, using population-based registry data, are considered to be a better reflection of the “real world” [5,23]. However, although large study populations can be derived from registries, the observational design means that confounding by indication must be considered when studying treatment effects.

A limitation of our study was that the selected populations differed in some respects. Advanced stage and higher grade cases were more frequently observed in Ireland. Although the analyses included patients aged 65 and older, this finding might be explained partly by differences in screening [24,25] and possibly methods of grading between countries. To overcome the difference in stage distribution, we grouped all analyses by stage. Slight under-ascertainment of radiotherapy treatments is known to have occurred among Irish patients who had breast surgery in private hospitals. However, only about 17% of surgical patients in the age 65+ group falls into this category, and we estimate that the percentages of Irish patients reported as having radiotherapy in Ireland may about 2% too low, not enough to affect our conclusions.

To achieve best practice for older breast cancer patients, possibly, attention should be shifted to other outcomes rather than survival to improve quality of care for older breast cancer patients. However, we could draw no conclusions on aspects such as quality of life, risk of recurrence or complications, as we did not have data on these aspects. In addition, because of full anonymization of the datasets used for our analysis, characteristics of hospitals, such as the type (academic/teaching hospital, private/public clinic), but also the presence of radiotherapy facilities were not available. Therefore, we were unfortunately not able to see if guideline-adherence was associated with hospital characteristics.

The retrospective design of the current study, despite the positive arguments mentioned previously, remains a limitation. However, because of the availability of comprehensive cancer registry data, it was possible to create a large database of population-based, generalizable data.

In the future, study designs in which countries are compared on treatment strategy and breast cancer outcome are likely to be applied more frequently. By including many countries in analyses, specific populations that differ on only one treatment modality could be identified. Consequently, more evidence can be obtained from observational studies, by comparing patient outcomes between countries using an instrumental variable study design [26].

The European Registration of Cancer Care, or in short European Cancer Audit (EURECCA) aims [27] to create a population-based audit structure that covers all breast cancer patients across Europe: anonymous patient and tumor data, including treatment and outcome information will be registered in a uniform way across countries. The aim is to develop an extensive data source with the ultimate goal to define high-quality care and monitor the quality of care of all European cancer patients and so improving outcome of cancer care. EURECCA aims to investigate best practices and learn from them, as well as perform analysis on patient groups that deviate from guidelines such as the young and elderly. The availability of comprehensive cancer registry data, (like that used in the current study) facilitates the identification of large cohorts of population-based, generalizable data.

In conclusion, in this population-based study comparing patterns of care and survival of older breast cancer patients on a national scale in The Netherlands and Ireland, we found large differences in treatment approach, with more guideline-adherence on locoregional treatment in The Netherlands, and more prescription of systemic therapy in Ireland. Patients in Ireland had a worse relative survival as compared with the Dutch patients, although it was not possible to link this survival difference directly to differences in one or more of the specific treatment modalities. However, our finding should be a strong recommendation to perform more research on an international scale, with the ultimate goal to equalize the survival rates for breast cancer patients across Europe.

Supporting Information


The authors thank the registration teams of the Comprehensive Cancer Centre Netherlands, Comprehensive Cancer Centre South and the Irish National Cancer Registry for the collection of data for the Netherlands and Irish National Cancer Registries, as well as the scientific staff of the Comprehensive Cancer Centre Netherlands for scientific advice. The results of this study were partly presented at the conference of the International society of geriatric oncology (SIOG), October 2012, Manchester, United Kingdom.

Author Contributions

Conceived and designed the experiments: M. Kelly PW EB GJL. Analyzed the data: M. Kelly PW EB MBK. Contributed reagents/materials/analysis tools: M. Kiderlen PW. Wrote the paper: M. Kiderlen PW EB M. Kelly RA PB CB OD AC CV GJL.


  1. 1. Wildiers H, Kunkler I, Biganzoli L, Fracheboud J, Vlastos G, et al. (2007) Management of breast cancer in elderly individuals: recommendations of the International Society of Geriatric Oncology. Lancet Oncol 8: 1101–1115. pmid:18054880
  2. 2. Biganzoli L, Wildiers H, Oakman C, Marotti L, Loibl S, et al. (2012) Management of elderly patients with breast cancer: updated recommendations of the International Society of Geriatric Oncology (SIOG) and European Society of Breast Cancer Specialists (EUSOMA). Lancet Oncol 13: e148–e160. pmid:22469125
  3. 3. Townsley CA, Selby R, Siu LL (2005) Systematic review of barriers to the recruitment of older patients with cancer onto clinical trials. J Clin Oncol 23: 3112–3124. pmid:15860871
  4. 4. van de Water W, Kiderlen M, Bastiaannet E, Siesling S, Westendorp RG, et al. (2014) External validity of a trial comprised of elderly patients with hormone receptor-positive breast cancer1. J Natl Cancer Inst 106: pmid:25535299
  5. 5. Al-Refaie WB, Vickers SM, Zhong W, Parsons H, Rothenberger D, et al. (2011) Cancer trials versus the real world in the United States. Ann Surg 254: 438–442. pmid:21775882
  6. 6. Mieog JS, de Kruijf EM, Bastiaannet E, Kuppen PJ, Sajet A, et al. (2012) Age determines the prognostic role of the cancer stem cell marker aldehyde dehydrogenase-1 in breast cancer. BMC Cancer 12: 42. pmid:22280212
  7. 7. Vandenbroucke JP (2004) When are observational studies as credible as randomised trials? Lancet 363: 1728–1731. pmid:15158638
  8. 8. Altman DG, Bland JM (1999) Statistics notes. Treatment allocation in controlled trials: why randomise? 4. BMJ 318: 1209.
  9. 9. Kiderlen M, Bastiaannet E, Walsh PM, Keating NL, Schrodi S, et al. (2012) Surgical treatment of early stage breast cancer in elderly: an international comparison. Breast Cancer Res Treat 132: 675–682. pmid:22119939
  10. 10. Greene FL, Page DL, Fleming ID (2002) AJCC cancer staging manual, 6th edition. New York: Springer.
  11. 11. Fleming ID, Cooper JS, Henson DE, Hutter RVP, Kennedy BJ, et al. (1997) AJCC Cancer Staging Manual, fifth edition. Philadelphia, PA: Lippincott-Raven.
  12. 12. CBO (2005) Richtlijn Behandeling van het mammacarcinoom 2005.
  13. 13. O’Higgins N, Health Information and Quality Authority (2007) National Quality Assurance Standards for Symptomatic Breast Disease Services—Developing Quality Care for Breast Services in Ireland.
  14. 14. Walsh TN, O’Higgins N, Royal College of Surgeons in Ireland (2000) Breast Cancer Management Clinical Guidelines.
  15. 15. Hakulinen T, Seppa K, Lambert PC (2011) Choosing the relative survival method for cancer survival estimation. Eur J Cancer 47: 2202–2210. pmid:21549589
  16. 16. Dickman PW, Sloggett A, Hills M, Hakulinen T (2004) Regression models for relative survival. Stat Med 23: 51–64. pmid:14695639
  17. 17. Allemani C, Storm H, Voogd AC, Holli K, Izarzugaza I, et al. (2010) Variation in ‘standard care’ for breast cancer across Europe: a EUROCARE-3 high resolution study. Eur J Cancer 46: 1528–1536. pmid:20299206
  18. 18. Garcia-Etienne CA, Tomatis M, Heil J, Friedrichs K, Kreienberg R, et al. (2012) Mastectomy trends for early-stage breast cancer: a report from the EUSOMA multi-institutional European database. Eur J Cancer 48: 1947–1956. pmid:22483323
  19. 19. van Nes JG, Seynaeve C, Jones S, Markopoulos C, Putter H, et al. (2010) Variations in locoregional therapy in postmenopausal patients with early breast cancer treated in different countries. Br J Surg 97: 671–679. pmid:20393978
  20. 20. Allemani C, Sant M, Weir HK, Richardson LC, Baili P, et al. (2012) Breast cancer survival in the US and Europe: A CONCORD high-resolution study. Int J Cancer. pmid:23233329
  21. 21. van de Water W, Bastiaannet E, Dekkers OM, de Craen AJ, Westendorp RG, et al. (2012) Adherence to treatment guidelines and survival in patients with early-stage breast cancer by age at diagnosis. Br J Surg 99: 813–820. pmid:22492310
  22. 22. Sullivan P, Goldmann D (2011) The promise of comparative effectiveness research. JAMA 305: 400–401. pmid:21266687
  23. 23. Dreyer NA, Garner S (2009) Registries for robust evidence. JAMA 302: 790–791. pmid:19690313
  24. 24. RIVM (2012) Bevolkingsonderzoek Borstkanker.
  25. 25. National Cancer Screening Service (2012) The National Breast Screening Program.
  26. 26. Rassen JA, Brookhart MA, Glynn RJ, Mittleman MA, Schneeweiss S (2009) Instrumental variables I: instrumental variables exploit natural variation in nonexperimental data to estimate causal relationships 3. J Clin Epidemiol 62: 1226–1232. pmid:19356901
  27. 27. (2014) EURECCA—European Registration of Cancer Care.