Childhood Cancer Substantially Underdiagnosed Worldwide

Zachary Ward, Harvard TH Chan School of Public Health

Zachary Ward

Childhood cancer is significantly underdiagnosed—especially in south Asia and sub-Saharan Africa—underscoring the need for stronger health systems and the expansion of universal health coverage, according to an analysis published in Lancet Oncology.¹

Using a Global Childhood Cancer microsimulation model, investigators showed that 43% (172,000/397,000) of all childhood cancers went undiagnosed in 2015. In 2015, there were approximately 397,000 (95% uncertainty interval [UI], 377,000-426,000) incident cases of childhood cancer worldwide, of which only 224,000 (95% UI, 216,000-237,000) were detected and diagnosed.

These annual estimates are comparable to those from the International Agency for Research on Cancer, which estimated 200,000 diagnosed cases in 2018, as well as the Global Burden of Disease Study, which approximated 195,000 diagnosed (95% UI, 175,000-206,000) cases in 2016.

Although accurate estimates of childhood cancer incidence are needed by policymakers to inform important health priorities and decisions, many countries do not have cancer registries in place to collect these necessary data. Even when such registries exist, undetected diagnoses could result in a significant underestimation of true incidence. As such, investigators aimed to provide precise estimates of total incidence of childhood cancer in which they account for underdiagnosis, in this analysis.

Based on trends in population growth and urbanicity, geographical disparities in cancer incidence, and confines within health systems that limit access and proper referral to cancer care, investigators developed a microsimulation model to calculate the incidence of childhood cancer among 200 countries and territories around the world.

Investigators stratified these countries and territories into 4 income brackets per the World Bank and 21 geographical regions per the United Nations (UN). To properly account for accurate cancer diagnosis, investigators charted a presumed chain of events including access to primary care, referral, and ultimate diagnosis. Access was modeled on an index of antenatal care coverage, vaccination coverage, and the World Health Organization composite coverage indicator, which includes reproductive, maternal, newborn, and child health interventions.

Subsequently, referral was modeled after various indicators for appropriate treatment of an illness, such as potential pneumonia and antidiarrheal treatment with oral rehydration salts or therapy.

To ensure the model’s accuracy, investigators calibrated the model to public cancer registry data in children aged 0 to 14 years, in which the observed data were fixed, and cancer incidence and probabilities of health system access and referral served as random variables.

Registry-reported cancer cases were collected from the International Incidence of Childhood Cancer. Among the 200 total provinces included in the study, registry data from 77 countries were incorporated into the model.

“Our model suggests that nearly 1 in 2 children with cancer are never diagnosed and may die untreated,” said lead author Zachary Ward, a doctoral student in health policy at Harvard TH Chan School of Public Health, in a recent press release.² “This new model provides specific estimates of childhood cancer that have been lacking.”

In addition to the significant underestimation of incident cases of childhood cancer in 2015, the model found significant disparities by region, ranging from 3% in western Europe (120 of 4300) and North America (300 of 10,900) to 57% in western Africa (43,000/76,000). Further, among the overall population of estimated cases in south Asia, approximately 49% (67,000/137,000) were estimated to be undiagnosed. Moreover, 92% (366,000/397,000) of the total incident cases of childhood cancer were observed in low- and middle-income countries.

Specified diagnoses included acute lymphoblastic leukemia (ALL), non-Hodgkin lymphoma, nephroblastoma, Burkitt lymphoma, retinoblastoma, acute myeloid leukemia, Hodgkin lymphoma, gangloneuroblastoma, astrocytoma, central nervous system embryonal tumors, rhabdomyosarcoma, osteosarcoma, Kaposi sarcoma, gonadal germ cell carcinoma, and Ewing and related cancers.

With the exception of sub-Saharan Africa, ALL was the most common malignancy in most geographical regions. However, investigators noted that a higher incidence of other malignancies led to a higher overall cancer incidence in Africa. In western Africa, the estimated age-standardized total incidence rate was 430 (95% UI, 344-546) per million person-years, whereas 157 (95% UI, 151-161) per million-person years was the average in Europe and North America.

Approximately 75% of this difference in total incidence was attributable to a higher proportion of lymphomas, retinoblastoma, and renal tumors in western Africa—a trend that is likely to be compounded by estimates of substantial population growth.

After accounting for population growth in each country using the UN probabilistic projections, investigators predicted that there will be 6.7 million (95% UI, 6.3-7.2 million) cases of childhood cancer worldwide from 2015 to 2030. In countries without UN projections, investigators input projections according to regional trends. To account for potential projection errors, investigators built in uncertainty intervals around their projections of percentage change in the population size based on the regional average.

Based on existing healthcare systems, investigators approximated that 2.9 million (95% CI, 2.7-3.3) of these cases will go undiagnosed during that period.

“Health systems in low-income and middle-income countries are clearly failing to meet the needs of children with cancer,” said senior study author Rifat Atun, professor of global health systems at Harvard University, director of Global Health Systems Cluster at Harvard TH Chan School of Public Health. “Universal health coverage, a target of UN Sustainable Development Goals, must include cancer in children as a priority to prevent needless deaths.”

Irrespective of the model’s statistical limitations in terms of proxy indicators and lack of registry data, investigators were able to provide the first global quantification of past and future estimates of childhood cancer incidence. Such a model is crucial in informing future healthcare policy that is better equipped to diagnose and care for children with cancer, the authors concluded.

References

1. Ward ZJ, Yeh JM, Bhakta N, et al. Estimating the total incidence of global childhood cancer: a simulation-based analysis [published online February 26, 2019]. Lancet Oncol. doi: 10.1016/S1470-2045(18)30909-4.
2. Nearly half of childhood cancers worldwide undiagnosed [news release]. Boston, MA: Harvard TH Chan School of Public Health. Published February 26, 2019. https://bit.ly/2C0SPBs. Accessed March 1, 2019.