Microsimulation model with a macro impact

Science Spotlight

Microsimulation model with a macro impact

From the Etzioni Group, Public Health Sciences Division

Sept. 17, 2018

The global cancer burden will to continue in an upward trend in coming decades with disproportionate increases in low-income and middle-income countries (LMICs). As the most common cancer among women, breast cancer survival rates vary substantially among various regions worldwide. For example, the five-year survival rate in the United States is nearly 90% but in some regions in Africa it is less than 50%. Such geographic variation in survival rates for certain types of cancer, including breast cancer, is at least partially due to late stage presentation and inadequate treatment options. Modifications in the allocation of limited health care resources could improve breast cancer outcomes for women in LMICs, but ministries of health require guidance on which interventions could potentially yield the best outcome improvements. New results from a microsimulation study of changes in breast cancer detection and treatment strategies addressed this question and were recently published in Lancet Global Health by Dr. Ruth Etzioni and researchers in the Division of Public Health Sciences. Microsimulation, or microanalytic simulation, is a computational analytic method for estimating the potential impact of variables or interventions on outcomes in large populations.

Multiple strategies can be implemented to improve breast cancer survival rates in LMICs. Mammographic screening is the primary method for early detection in high income countries with highly evolved healthcare infrastructure but is unaffordable in most of the world. Furthermore, women in LMICs commonly present with clinically obvious late stage disease where image detection is unnecessary to detect the tumor’s presence. Thus, in limited resource settings, downstaging of cancer through improved clinical detection (awareness education, clinical breast examination and prompt tissue diagnosis) is a necessary prerequisite to organized mammographic screening and can itself improve outcome. At the same time, multimodality treatment including systemic (drug) treatment is required to reduce the risk of metastatic progression of disease and improve breast cancer survival. Early diagnosis without effective treatment cannot be anticipated to shift survival curves.

To decide which individual or combined strategies should be implemented in their setting, policy makers require evidence-based guidance on the projected relative impact of early detection and treatment programs in their unique setting. In the new study, the authors focused specifically on these two main strategies and conducted a microsimulation study to estimate the magnitude by which the implementation of each strategy, independently or in combination, could potentially impact breast cancer mortality among women in LMICs. This is the first microsimulation tool to permit the simultaneous assessment of early diagnosis and improved treatment strategies applied to varying degrees in the same limited-resource population.

For low-income countries, the authors modeled sub-Saharan Eastern Africa, while for middle-income countries they modeled Colombia. For each region, breast cancer records were generated for a hypothetical population. For the fraction of women in each simulated population, the authors set major factors that influence prognosis, including: age, cancer stage at diagnosis, and whether the cancer was estrogen-receptor (ER) positive or negative. Three different detection scenarios, which would influence distribution of stage at diagnosis, were included in the model: poor breast cancer awareness with no screening, increased awareness and clinical screening, and screening with mammograms. In addition, four different treatment programs (A through D) were included in the model: current standard-of-care (program A), endocrine therapy for ER-positive cases (program B), program B plus chemotherapy for ER-negative cases (program C), and program C plus chemotherapy for advanced ER-positive cases (program D).


Graphical representation of absolute risk reduction for each intervention from the microsimulation model.

Absolute risk reduction for each intervention compared with the standard of care in the East African region. Percentages in parentheses represent the proportions of women who present with advanced-stage disease. B=endocrine therapy for ER-positive cases. C=B plus chemotherapy for ER-negative cases. D=C plus chemotherapy for advanced ER-positive cases. Error bars are 95% uncertainty intervals.

Image from the publication

Ten different possible policies were generated as combinations of the three detection scenarios and treatment programs and then modeled for each geographical region. To determine how the policies fared against one another, three outcome measures were generated for each: mortality rate ratio (relative reduction in mortality), absolute risk reduction (number of lives saved), and years of life saved.

Compared to the current standard-of-care, the simulated treatment and detection strategies would save 23 (no early detection with treatment program B) to 113 (elevated awareness with clinical screening with treatment program D) lives in the East Arica region (see Figure). The corresponding relative reduction in mortality ranged from 8 – 41% and a range of 92 – 418 years of life saved. The authors noted that the greatest benefit in the estimated outcomes occurs in the scenario of no early detection, with a range of 23 – 60 lives saved compared to the detection scenario of clinical screening, with a range of 87 – 114 lives saved. These results demonstrate that with improvements in early cancer detection, the benefit specifically attributed to treatment are less pronounced.

Similar general trends were observed in the simulation of the Colombian cohort. Among the policies, the number of lives saved ranged from 31 (no early detection with treatment program B) 105 (mammographic screening with treatment program D) with a relative reduction in mortality of 7 – 25%.

Globally, cancer is the cause of one in seven deaths. With better screening procedures, translating into earlier detection, and improved treatment therapies, this study demonstrates that breast cancer mortality rates could be significantly reduced and years of lives saved increased in LMICs. The authors have made the newly developed, user-friendly microsimulation model available online. The model may be a useful tool in guiding LMICs in the development of cancer-related policies in which adjustments in the allocation and use of health care resources available may maximize their potential to reduce cancer mortality.


This research was supported by the National Institutes of Health and the Susan G. Komen Foundation.

Research reported in the publication is a collaboration between Cancer Consortium members Ruth Etzioni (Fred Hutch) and Benjamin Anderson (UW).

Birnbaum JK, Duggan C, Anderson BO, Etzioni R. 2018. Early detection and treatment strategies for breast cancer in low-income and upper middle-income countries: a modelling study. Lancet Glob Health. doi: 10.1016/S2214-109X(18)30257-2