Defining the local incidence of cancer is a key first step towards developing a comprehensive cancer control strategy [1]. In the context of childhood cancer, such information helps to understand disease etiology, improve access to care, plan investments in service delivery, advocate resource allocation, and measure the quality of different components of the health system [1].

Estimates of global and country-specific cancer and childhood cancer burden are provided by multiple groups. The recently published GLOBOCAN 2018 study [2], coordinated by the International Agency for Research on Cancer, provides comprehensive global childhood cancer incidence estimates and is commonly used by the World Health Organization and governments for planning cancer control. In 2018, the study estimated that 200,166 new children, age 0-14 years, were diagnosed and registered with cancer globally, of whom 28,712 (14.3%) were from India [2].

Recent analyses have questioned the accuracy of GLOBOCAN data for estimating the incidence of childhood cancer [3]. The local incidence of childhood cancer varies substantially in the published data including that from India [4,5]. It has been hypothesized that under-diagnosis and consequently under-registration, which is disproportionately high in low and middle income countries (LMIC), leads to an "incidence gap" and under-estimates the cancer burden, and are hence not reflected in the GLOBOCAN 2018 data [6]. This theory has been further substantiated by independent simulation-based studies that have estimated the annual global childhood cancer burden is nearly 45% greater than that historically reported, between 360,000 to 400,000, when children who develop cancer but are never registered are counted [7,8].

Due to perceived incomplete case-finding, misdiagnosis within the fragmented Indian health system and significantly lower incidence-rates of childhood cancer in India, the currently reported childhood cancer from GLOBOCAN 2018 likely represent an under-estimate [5,9]. In this study, we aim to use observed data rather than simulation to estimate the number of children (0-14 years), as well as number of children and adolescents (0-19 years), in India who develop cancer every year. Additionally, we report these data at the national and state/union territory (UT) level for the purposes of supporting cancer control planning.

Age-specific (five year groups), sex-specific, and state/UT- specific population data from India Census 2011 was used [10]. These data pre-date the division of Andhra Pradesh in 2014 and Jammu and Kashmir in 2019 and hence considers these states as a whole. Conducted every 10 years since 1872, phase one of the 2011 census began on 1st April 2010 and included house-listing and collecting information for the National Population Register. The second phase was the population enumeration phase done from 9 to 28 February, 2011.

Global average incidence rates from the International Incidence of Childhood Cancer 3 (IICC3) report were used [4]. Conducted by the International Agency for Research on Cancer with the specific purpose of collecting and disseminating childhood cancer data, IICC-3 is the third monograph following from IICC-1 published in 1988 and IICC-2 published in 1998. Only population based cancer registries were invited. The target period covered the years starting with 1990, and targeted the age range of 0-19 years. IICC-3 uses observed data on cancer incidence from countries or regions covered by population-based cancer registries and unlike GLOBOCAN does not extrapolate to produce selected national, regional or global cancer burden estimates.

Incidence rates per million person-years for the 0-14 years (children) and 0-19 years (children and adolescents) age groups were age-adjusted using the world standard population to provide age-standardised incidence rates, using the age-specific incidence rates for individual age groups (0-4 years, 5-9 years, 10-14 years, and 15-19 years).

Statistical analyses: Number of incident cases for 0-14 years, 0-19 years and individual age groups (0-4 years, 5-9 years, 10-14 years, and 15-19 years) was calculated by multiplying incidence rates with the denominator population for the country and each state/UT. To get cancer-specific incident cases according to the International Childhood Cancer Classification third edition in 0-14 years age group, cancer-specific incidence rates were multiplied with the denominator population for the country and each state/UT [11]. As cancer-specific incidence rates were not available for 0-19 year age group, cancer-specific incident cases for this age group were obtained by adding incident cases in the 0-14 year age group derived above and cancer-specific incident cases in the 15-19 year age group. To derive the cancer-specific incident cases in 15-19 year age group, cancer-specific incidence rates for this agre group were multiplied with the denominator population for the country and each state/UT.

Table III  International Childhood Cancer Classification Type-Specific Incident Cases* of Cancer 
in Children 0-19 Years of Age in States and Union Territories of India

The National Cancer Registry Program (NCRP) in India provides data for the observed individual population based cancer registries which include all patients with cancer diagnosed and registered, and cover less than 10% of the Indian population [12]. The NRCP report, however, does not extrapolate to provide an estimate of the national incidence of childhood cancer. National estimates used for cancer control planning in India are provided by the GLOBOCAN 2018 models that are built using individual cancer registry data from the NCRP report, national vital statistic data sets and economic development covariates [2,12]. In this analysis, using internationally standardized incidence rates and population-estimates from India, we found that the incidence of childhood cancer is 54.8% larger in 0 to 14 years age range (52366 vs 28712) and 50.3% larger in 0 to 19 years age range (76805 vs 38640) compared to GLOBOCAN 2018. We hypothesize the large observed difference between the two estimates is due to the substantial number of cases that are not diagnosed and/or registered in India [6-8].

For health systems planning, calculating both the number of patients who will develop cancer and the number of patients who are diagnosed and registered is critical information. Knowing the current healthcare utilization needs presently is critical for states to make allocation decisions today. However, as cancer control plans typically are written as multi-year plans, identifying the gap between the observed and expected cases is important. In particular, as strategies to improve access and referral are often built into national cancer control plans, these calculations can inform prioritization, decision-making, monitoring procedures and budgeting.

Not only is the incidence of diagnosed and registered (GLOBOCAN 2018) approximately half of those who develop cancer (our estimates), Suppl. Table I shows this differential varies by age, gender and cancer. The estimated proportion of girls diagnosed and registered with cancer is 10% less than boys. This aligns with the narrative of female children with cancer experiencing relatively greater barriers to accessing healthcare [5,13-15]. Similarly the differential of the GLOBOCAN 2018 estimates and those from our analysis is greatest in CNS tumors and lowest in leukemias. This may reflect the relatively sick nature of leukemia patients, and easy availability of automated blood counts and bone marrow examination as compared to more sophisticated and technology dependent interventions like neuroimaging and neurosurgery. There is also a component of under-ascertainment in diagnosed CNS tumors as currently NCRP datasets exclude tumors with ‘benign’ or ‘uncertain’ behavior and such tumors constitute 40-50% of CNS tumors in children and adolescents [16].

Limitations of our analysis are that we are using the 2011 census data and hence have likely slightly over-estimated the incidence of new cases. Although the population of India is projected to peak around 2050, that for children ages 0-19 years is expected to peak between 2010 to 2020. And hence one can argue that the burden in 2011 will be higher by a few percentage points than the burden in 2020 and beyond. The census 2011 however remains the most reliable estimates of population at the state and union territory level and hence was used. It is also difficult to be more precise to the relative contributions of under-diagnosis versus under-registration although there is some evidence to support that under-diagnosis is the main component of ‘incidence gap’ in the burden [17]. The contribution of under-diagnosis and under-registration may vary across states depending on the healthcare accessibility but in our analysis we have assumed that it is same across states.

Perhaps the most important question in regard to our estimates is its reliability and accuracy. While there is a degree of uncertainty around the burden, its reliability can be inferred from two arguments. Firstly, is the central tenet that environment plays a minor role in the etiology of childhood cancer hence the variation in the incidence of childhood cancer across the world is limited [4,18]. Secondly, under-diagnosis and other aspects of impaired healthcare access like delayed diagnosis, abandonment of treatment, etc. are well-recognized issues in LMICs [5,14,17,19,20]. Our estimates of 45-50% under-diagnosed children mirrors other recently published data which reached similar conclusions using differing methodologies [7,8].

In conclusion, our analysis proposes new estimates of incident childhood cancer cases in India. We also provide estimates at state and union territory level. This has enormous implications for all childhood cancer stakeholders who aim to provide access, treatment and chance of long-term cure to every child with cancer. It also suggests that access to diagnosis is as big, if not a bigger problem, than access to complete treatment and needs to be tackled early and urgently. Future regional, national and international research on childhood cancer epidemiology and healthcare accessibility would help further refine these estimates.

1. World Health Organization. National Cancer Control Programs: Policies and Managerial Guidelines. World Health Organization; 2002.

2. International Agency for Research in Cancer. Global Cancer Observatory. Accessed February 3, 2020. Available from: http://gco.iarc.fr/

3. Bhakta N, Force LM, Allemani C, et al. Childhood cancer burden: A review of global estimates. Lancet Oncol. 2019;20:e42-e53.

4. Steliarova-Foucher E, Colombet M, Ries LAG, et al. International incidence of childhood cancer, 2001-10: A population-based registry study. Lancet Oncol. 2017;18:719-31.

5. Arora RS, Eden TO, Kapoor G. Epidemiology of childhood cancer in India. Indian J Cancer. 2009;46:264-73.

6. Howard SC, Lam CG, Arora RS. Cancer epidemiology and the "incidence gap" from non-diagnosis. Pediat Hematol Oncol J. 2018;3:75-8.

7. Johnston WT, Erdmann F, Newton R, et al. Childhood cancer: Estimating regional and global incidence. Cancer Epidemiol. 2020; :101662.

8. Ward ZJ, Yeh JM, Bhakta N, Frazier AL, Atun R. Estimating the total incidence of global childhood cancer: A simulation-based analysis. Lancet Oncol. 2019;20: 483-93.

9. Arora B, Kanwar V. Childhood cancers in India: Burden, barriers, and breakthroughs. Indian J Cancer. 2009;46: 257-9.

10. Office of the Registrar General and Census Commissioner, Ministry of Home Affairs, Government of India.Census of India 2011. Accessed Febuary 3, 2020. Available from: http://censusindia.gov.in/2011-Common/Census Data2011.html

11. Steliarova-Foucher E, Stiller C, Lacour B, Kaatsch P. International classification of childhood cancer. Cancer. 2005;103:1457-67.

12. National Centre for Diseases Informatics and Research, National Cancer Registry Programme, Indian Council of Medical Research. Three-year Report of Population Based Cancer Registries 2012–2014. Accessed May 17, 2020. Available from: http://ncdirindia.org/ NCRP/ALL_NCRP_ REPORTS/PBCR_ REPORT_2012_2014 /index.htm

13. Bhopal SS, Mann KD, Pearce MS. Registration of cancer in girls remains lower than expected in countries with low/middle incomes and low female education rates. Br J Cancer. 2012;107:183-8.

14. Hazarika M, Mishra R, Saikia BJ, et al. Causes of treatment abandonment of pediatric cancer patients – Experience in a regional cancer centre in North East India. Asian Pac J Cancer Prev. 2019;20:1133-7.

15. Bhargav A, Singh U, Trehan A, Zadeng Z, Bansal D. Female sex, bilateral disease, age below 3 years, and apprehension for enucleation contribute to treatment abandonment in retinoblastoma. J Pediatr Hematol Oncol. 2017;39:e249-53.

16. Arora RS, Alston RD, Eden TO, et al. Age-incidence patterns of primary CNS tumors in children, adolescents, and adults in England. Neuro Oncol. 2009;11:403-13.

17. Arora RS. Why is the incidence of childhood cancer lower in rural India? Cancer Epidemiol. 2010;34:105-6.

18. Stiller CA, Parkin DM. Geographic and ethnic variations in the incidence of childhood cancer. Br Med Bull 1996;52:682-703.

19. Friedrich P, Lam CG, Itriago E, et al. Magnitude of treatment abandonment in childhood cancer. PLoS One. 2015;10:e0135230.