Pediatric Hematology Oncology Journal 3 (2018) 79e87

Contents lists available at ScienceDirect

Pediatric Hematology Oncology Journal

journal homepage: https://www.elsevier.com/journals/pediatric-
hematology-oncology-journal/

Childhood cancer survival disparities in a universalized health system

in Cali, Colombia*
Oscar Ramirez a, b, c, *, Paula Aristizabal e, Alia Zaidi f, Anne Gagnepain-Lacheteau g,
Raul C. Ribeiro f, Luis E. Bravo a, d, On behalf of VIGICANCER working group
a
Cali's Cancer Population-based Registry, Universidad del Valle, Cali, Colombia
b
POHEMA (Pediatric Oncologists and Hematologist) Foundation, Cali, Colombia
c
Centro M
edico Imbanaco de Cali, Colombia
d
Department of Pathology, Universidad del Valle, Cali, Colombia
e
Department of Pediatrics, Division of Pediatric Hematology/Oncology University of California San Diego/Rady Children's Hospital, Reducing Cancer
Disparities Program University of California San Diego Moores Cancer Center, La Jolla, CA, USA
f
St Jude Children's Research Hospital, Memphis, TN, USA
g
Fondation Sanofi Espoir, France

a r t i c l e i n f o a b s t r a c t

Article history: Universal healthcare has been proposed as a strategy to achieve health equity. Herein, we describe
Received 7 May 2018 childhood cancer survival disparities within a universal healthcare system in Cali, the third largest city in
Received in revised form Colombia. We prospectively included data from Cali's childhood cancer surveillance system (VIG-
25 January 2019
ICANCER) cohort (2009e2016) and adjusted the hazard ratios (aHR) for confounders using multivariate
Accepted 29 January 2019
Available online 10 February 2019
Cox regression. We included 1808 patients with a median age for children (n ¼ 1499) of 6 years and for
adolescents (n ¼ 309) of 17 years. Fifty-six percent were male, 14% were afro-descendants, 61% resided
outside of Cali, 52% had public insurance, 44% had private insurance, and 4% were uninsured. Five-year
Keywords:
Health disparities
overall survival rates for patients with private insurance, public insurance and uninsured patients were
Universal health care coverage 62% (95% CI: 58, 66), 43% (95% CI: 39, 46) and 23% (95% CI: 13, 35), respectively. Compared to private
Survival insurance, mortality among patients with public insurance (aHR ¼ 1.6; 95% CI: 1.3, 1.9) and uninsured
Childhood (aHR ¼ 2.7; 95% CI: 1.9, 4.0) was higher. We found significant disparate survival outcomes, primarily by
Epidemiology insurance and tumor type. Higher treatment abandonment, higher treatment-related mortality, and
Cancer advanced disease at diagnosis partially explained these disparities. Survival inequalities persist in
Treatment outcome Colombia despite an established universal healthcare system aimed at providing equal care for all.
© 2019 Publishing Services by Elsevier B.V. on behalf of Pediatric Hematology Oncology Chapter of Indian
Academy of Pediatrics. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction preventable, cancer control in the childhood population is depen-

dent on the overall effectiveness of cancer services delivery and
Cancer in children is highly curable in approximately 80% of the health systems.
cases [1,2], but effective treatment is complex and costly, particu- Survival disparities in children with cancer vary widely within
larly in low and middle income countries (LMIC), where resources and between countries [4]. Cancer biology, host conditions, social
are constrained [3]. Moreover, as childhood cancer is currently not determinants, and the delivery of cancer care contribute to survival
disparities [5e7]. In high-income countries (HIC) ethnicity, race and
socioeconomic status (SES) have been associated with survival
*
Supported by the My Child Matters program of Sanofi-Espoir Foundation (O.R., disparities [5e7]. In LMIC, additional factors include delayed diag-
A.Z., A.G-L.) and the POHEMA Foundation (O.R. and L.E.B.); the American Lebanese nosis, lack of training in pediatric cancer management and support,
Syrian Associated Charities (R.C.R.).
limited access to curative care, cultural barriers, treatment aban-
* Corresponding author.Cali's Cancer Population-based Registry, Universidad del
Valle, Cali, Colombia. donment, poor adherence to treatment, and fragmented healthcare
E-mail address: [email protected] (O. Ramirez). delivery [8,9]. In Colombia, childhood cancer survival disparities
Peer review under responsibility of Pediatric Hematology Oncology Chapter of have not been reported at the population-based level.
Indian Academy of Pediatrics.

https://doi.org/10.1016/j.phoj.2019.01.001
2468-1245/© 2019 Publishing Services by Elsevier B.V. on behalf of Pediatric Hematology Oncology Chapter of Indian Academy of Pediatrics. This is an open access article
under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
80 O. Ramirez et al. / Pediatric Hematology Oncology Journal 3 (2018) 79e87

Table 1
Colombia, Valle del Cauca, and Cali demographic and social-economic indicators
Abbreviations
Colombia has an average health expenditure compared to other countries in the
region and 30% lower than the OECD average (21%). It is worth pointing that during
aHR Adjusted Hazard Ratio the five year study period there was no relative increase in health expenditure.
ALL Acute Lymphoid Leukemia During the study period Valle del Cauca province was wealthier than Colombian
CNS Central nervous system average, with a lower poverty percentage, but still with high economic disparities as
measured by GINI coefficient. Cali continues to be one of the five wealthiest capital
EFS Event free survival
cities in Colombia, showing improving in health indicators such as a steady decrease
ICD-O-3 International Classification of Diseases for in infant mortality. GINI coefficient is a measure of wealth distribution of a nation's
Oncology 3rd version residents; commonly used measurement of inequality.
ICCC-3 International classification of childhood cancer
Indicator Year
3rd version
2009 2016
CI confidence interval
LMIC Low and middle income countries Colombia
HR Hazard ratio Population 44 978 832 48 747 708
Population <19 years of age 16 593 426 16 317 781
OS Overall survival Per capita gross national income 5090 6310
POSc Contributory regimen GINI coefficient 0.56 0.52
POSs Subsidized regimen Poverty (%) 40.3 28
PPNA Population with no insurance Total expenditure on health (%) 6.4 6.2
Infant mortality rate (per 1000 live births) 18.8 16.8
POU Pediatric Oncology Unit
Province of Valle del Cauca
SES Socio-economic status Population 4 338 216 4 660 741
VIGICANCER Cali's childhood cancer surveillance system Population <19 years of age 1 454 089 1 393 429
GINI coefficient 0.52 0.48
Poverty (%) 33.3 22.6
Infant mortality rate (per 1000 live births) 13.7 11.5
Province's capital city Cali
Children with cancer deserve equitable, high quality care. Uni-
Population 2 219 720 2 394 925
versal health coverage has been proposed as a strategy to achieve Population <19 years of age 716 514 696 779
health equity (defined in terms of equal use for equal need) in GINI coefficient 0.50 0.48
different countries [10e12]. Under this premise, Colombia estab- Poverty (%) 28.4 15.4
Infant mortality rate (per 1000 live births) 9.7 7.8
lished a compulsory universal health insurance system in 1993
[13e16]. This system improved universal access to healthcare, Sources: Colombian national bureau of statistics and Cali's planning office [51,52].
diagnostic resources, therapeutic technologies and availability of
chemotherapy and other medications. Additionally, financial risks
associated to catastrophic diseases, such as cancer, have been Obligatorio de Salud Subvencionado), which covers approximately
minimized, and human resources training has been optimized, as 48% of the population. A minor proportion (<15%) falls outside of
demand for services has increased inmensely [13e16]. However, these two sub-systems: those with private insurance in addition to
operational barriers had been described in the public insurance POSc, those with government insurance (police, military, or gov-
system, such as lack of institutional capacity, deficient operational ernment employees), and those who are uninsured (PPNA, Pobla-
management, lacking information systems, and delays in the flow cion Pobre No Asegurada). Third-party private and public
of resources [15]. companies are responsible of client affiliation and access to ser-
The introduction of this universal health system resulted in an vices. Insurance companies receive government funding based on
initial increase of the overall survival (OS) of children with cancer in the number of individuals insured, and payments by private em-
Cali, the third largest city in Colombia, from 31.6 ± 3.2% ployers to support the contributory sub-system. Currently, both
(1992e1996) to 54.9 ± 2.8% (2002e2007) [16]. However, Cali's sub-systems cover basic services such as diagnosis work-up,
recent survival estimates suggest stagnation, with OS of 52% for the treatment, and rehabilitation of children with cancer.
2009e2013 cohort. In the last decade in Cali, important changes in the management
Herein, we describe childhood cancer survival disparities within of childhood cancer have occurred. The most relevant is a patient
the Colombian healthcare system and the burden of childhood drift from the public hospital to two large private pediatric
cancer mortality to help public health authorities address in- oncology units (POUs). In 2009, the public hospital POU1 served
equalities in care and outcomes in our population and in other 48% of children and adolescents with cancer in the region and in
similar resource-constrained settings. 2016 served only 11% of children. In 2016, the two large private
POUsb, c served 90% of the population. Cali had five pediatric on-
cologists in 2009, and this number increased to eight in 2013.
2. Methods
Before 2011, adolescents with cancer were treated at adult oncology
units. In 2012, it was mandated by the ministry of health that ad-
2.1. Healthcare system in Colombia and Cali
olescents needed to be treated at POUs.
Colombia is categorized as an upper-middle- income country
(Table 1), with a relatively young (1993- present) compulsory 2.2. Study cohort and measurements
universal health insurance system [13,15,16,18]. This system
[15e17,19] provides coverage to the majority of the population In 2009, we established a Childhood Cancer Outcomes Surveil-
through two different sub-systems: the contributory system for lance System (VIGICANCER) in Cali, with the support of Sanofi-
employed individuals, co-financed by the employer and the Espoir Foundation's “My Child Matters” program and Cali's
employee (Spanish acronym POSc, Plan Obligatorio de Salud Con- Population-based Cancer Registry [18,20]. VIGICANCER's rationale,
tributivo), which covers approximately 40% of the population and methodology and implementation have been previously published
the subsidized system (subsidized by the government) for informal [18]. VIGICANCER collects data on children and adolescents with a
employees and the unemployed (Spanish acronym POSs, Plan new diagnosis of cancer, who are initially treated in Cali, even if
 O. Ramirez et al. / Pediatric Hematology Oncology Journal 3 (2018) 79e87 81

they do not reside in Cali. VIGICANCER allows registration of pa- because the time to the event could not be estimated. The latest
tients from all POUs in Cali and monitoring of their clinical out- time point to follow-up was documented from the alive patients
comes. VIGICANCER collects baseline demographic variables (date with no events and considered as a censored observation. Cases of
of birth, sex, race/ethnicity, place of residence, place of diagnosis, treatment abandonment were included in the analyses as events
type of health insurance, and whether care is transferred to another and not as censored observations. Kaplan-Meier survival curves
institution) and clinical variables (pathological diagnosis, tumor were generated for EFS and OS from these survival probabilities,
location, date of diagnosis, and method of diagnosis). Transfer of and 95% confidence intervals (CIs) were calculated. We used Cox
care usually is dictated by insurance contracts with the POU. proportional hazards regression to estimate survival hazard ratios
Cancer types were coded using the International Classification (HRs), adjusted hazard ratios (aHR) and associated 95% CIs to
of Diseases for Oncology, 3rd edition (ICD-O-3) [21]. For hemato- evaluate the impact of health insurance on EFS and OS, adjusted for
logical and lymphoid tumors we used the World Health Organi- the following potential confounders: age, sex, afro-descendant
zation version of ICD-O-3 [22]. The International Childhood Cancer ancestry [5,6], place of residence [25], year of diagnosis, treat-
Classification version 3 (ICCC-3) was used for analyses [23]. Heath ment facility (private or public) [25] and care transferred to another
insurance information was obtained directly from patient admis- institution. For extracranial and lymphoid tumors, stage [26] was
sion forms, medical records or pathological reports. Insurance in- included in the model as a binary variable as metastatic vs. non-
formation was then classified in one of the four insurance metastatic. Collinearity was explored and was not found among
categories (sub-systems) of the healthcare system. For analysis the independent variables included in the models. We evaluated
purposes in this report, health insurance type was grouped into: a). the presumption of proportional hazards using Scho €enfeld re-
private insurance (POSc and private); b). public insurance (POSs siduals and logarithmically transformed survival time. No viola-
and government insurance): and c). uninsured (PPNA). VIG- tions of the assumption were observed. All P values reported were
ICANCER was approved by the Institutional Review Board at the 2-sided, and those that were 0.05 were considered to be statis-
University of the Valley and by all local POUs. tically significant.

2.3. Outcomes and follow-up

3. Results
The primary outcome was death. Secondary outcomes included
the following events: disease relapse, progression, development of 3.1. Cohort characteristics
second malignancy, and treatment abandonment. Relapse was
defined as histopathological evidence of disease recurrence after The age-standardized annual incidence rate of childhood cancer
previously documented complete remission. Diagnosis of relapse in Cali from 1977 to 2011 was 141 cases per million [17]. Our initial
was made without pathologic confirmation in cases where un- sample included a total of 1927 cases. Fifty patients (3%) had
equivocal evidence of relapse was available on imaging and/or missing data on health insurance type and were excluded from the
positive tumor markers. Disease progression was defined as at least analysis. Of those, 97% were followed-up using VIGICANCER during
25% increase in 2-dimension measurements of the visible tumor(s) the observation period 2009e2016. Therefore, analyses included
on relevant imaging. Second malignancy was defined as a histo- 94% (1808 cases) of the patients included in the larger VIGICANCER
logically distinct second cancer that developed after the first cohort. Of those, 1499 were children (<15 years of age) and 309
diagnosis. Treatment abandonment was defined as an unplanned were adolescents (15e18.9 years of age) [Fig. 1]. Ninety-six percent
delay of more than 4 weeks of the scheduled date to receive of case diagnoses had pathological confirmation.
curative intent therapy, that was unrelated to medical reasons and Median age in children and adolescents at diagnosis was 6 years
due to the patient not returning for care [24]. All cases of treatment (interquartile range [IQR]: 3, 11), and 17 years (IQR: 16, 18),
abandonment were included in VIGICANCER. Treatment abandon- respectively. Fifty-six percent (95% CI: 54, 59) were male, 14% (95%
ment was considered an event for survival analyses and docu-
mented in the same manner as no subsequent relapse or death.
1927 children and adolescents registered
VIGICANCER's clinical monitors gathered information every three
months from medical records and by contacting families to docu-
by VIGICANCER (2009 – 2016)
ment disease status (active vs. remission) and any information on
transfer of care between POUs. Lost to follow-up was defined as a
patient being non-contactable by any means for 6 months despite
69 lost to follow-up
of at least three phone contact attempts. After loss to follow-up was
documented, vital status of the patients was obtained from health-
1858 in active follow-up
insurance affiliation information, death certificates, and hospital
discharge registries (“passive surveillance”) through Cali's
Population-Based Cancer Registry. 50 without social security
data
2.4. Statistical analysis
1808 included in analyses
The study period was from January 1, 2009 to December 31,
2016, with an established last follow-up as December 31, 2017.
Statistical analyses were carried out using STATA (version 13.1;
StataCorp, College Station, TX). Event-free survival (EFS) was
calculated from the date of diagnosis to the date of first event.
Overall survival (OS) was calculated from the date of diagnosis to 790 950 68
date of death or treatment abandonment if no vital status was Private Public
known after this event. Patients detected only by death certificate
Uninsured
(without diagnosis date) were excluded for survival analyses, Fig. 1. Patient selection flow diagram.
82 O. Ramirez et al. / Pediatric Hematology Oncology Journal 3 (2018) 79e87

CI: 13, 16) were afro-descendants, and 61% (95% CI: 59, 63) resided patients with private insurance (35%) with an odds ratio (OR) of 1.9
outside of Cali. Distribution of solid vs. hematological tumors was (95% CI: 1.4, 2.7). Malignant bone tumors (OR 5.1; 95% CI: 1.9, 13.2)
similar (n ¼ 907 vs. n ¼ 901, respectively). and soft tissue sarcomas (OR 3.9; 95% CI: 1.1, 13.6) showed the
Overall, the major ICCC tumor groups were I (39%; 95% CI: 37, strongest association.
42), III (18%; 95% CI: 16, 19), and II (11%; 95% CI: 9, 12). The most As of December 31, 2017, 33% (95% CI: 35, 39) of patients had
frequent extra-cranial tumor in children (18% of all extra-cranial died. Proportion of deaths in hematological tumors and solid tu-
tumors [n ¼ 463]) and adolescents (40%, [n ¼ 162]) was bone tu- mors were 35% (95% CI: 34, 39) and 40% (95% CI: 31, 37), respec-
mor (ICCC group VIII). Thirty-six percent of patients with solid tively. Table 3 describes causes of death by health insurance type in
extracranial tumors had metastatic disease at diagnosis. hematological and solid tumors. Overall, 2% (95% CI: 1, 2) of the
Transfer of care among different POUs was frequent (33%; 95% patients died prior to starting cancer treatment. Death for Mortality
CI: 30, 35); with 53% (95% CI: 48, 57) being transferred among during induction therapy in hematological tumors was 5% (95% CI:
private POUs, 42% (95% CI: 37, 46) from public to private POUs, and 4, 7) and significantly higher in uninsured patients and in those
6% (95% CI: 4, 8) to POUs outside Cali. with public insurance compared to those with private insurance
(15% vs. 6% vs. 2%, respectively; P ¼ 0.015). Among patients with
3.2. Disparities in outcomes by health insurance acute lymphoblastic leukemia (ALL), treatment-related post-in-
duction mortality was 9% vs. 4% in patients with public and private
Patient characteristics by health insurance type are described in insurance, respectively (P ¼ 0.04). Ten patients (0.6%; 95% CI: 0.3,
Table 2. During the study period, 52% of the patients had public 1.0) died due to a second malignancy. Treatment-related mortality
insurance, 44% had private insurance and 4% were uninsured. Fifty was lower for solid than for hematological tumors (2% vs. 8%;
percent of patients residing in Cali had private insurance (95% CI: P < 0.001).
47, 54) compared to the public and uninsured groups combined Five-year cumulative incidence of relapse was 31% (95% CI: 28,
(30%; 95% CI: 27, 33). Twenty-one percent of patients with public 36) and it was lower for patients with private insurance (26%; 95%
insurance (95% CI: 18, 24) were transferred from a public POU to a CI: 22, 32), higher in patients with public insurance (36%; 95% CI:
private POU and 0.2% of patients were transferred from a private 30, 43), and highest in uninsured patients (54%; 95% CI: 30, 99).
POU to a public POU. Two-year cumulative incidence of treatment abandonment in
Metastatic extracranial solid tumors were more frequently children was 12% (95% CI: 7, 14). Treatment abandonment in pa-
diagnosed in patients with public insurance (65%) compared to tients with private insurance was 3% (95% CI: 2, 5), in patients with

Table 2
Characteristics of patients by health insurance type
Since 2009, there has been a continued decrease in uninsured patients as health system coverage improves. Private vs public insurance ratio was relatively stable during this
period, and only 18% of all patients were exclusively treated in the public POU. Public insurance was twice more frequent compared to private from those patients coming for
cities different of Cali; otherwise Cali patients were almost twice more frequently private than public. Pathology confirmation was evenly distributed by insurance status. Less
frequency of infant cases were observed in the public insurance and uninsured group, which is noteworthy as demographically those groups have mayor proportion of children
compared to private group population. Ratio of hematological tumors to solid tumors was similar among the three groups. Number of cases diagnosed and treated each year
was very similar.

Characteristics Health Insurance Category P-value

Private Public *None Total

n (%) n (%) n (%) n (%)

Age group (years)

[0,1) 62 (8) 45 (5) 2 (3) 109 (6) 0.001
[1, 5) 226 (29) 257 (27) 28 (41) 511 (28)
[5,10) 180 (23) 229 (24) 19 (28) 428 (24)
[10,15) 179 (23) 257 (27) 15 (22) 451 (25)
[15,19) 143 (18) 162 (17) 4 (6) 309 (17)
Total 790 (44) 950 (53) 68 (4) 1808 (100)
Male sex 446 (56) 542 (57) 36 (53) 1024 (57) 0.797
Afro-descendants (n ¼ 1527) 86 (13) 120 (15) 13 (25) 219 (14) 0.058
Place of residence (n ¼ 1806)
Province's capital city with POU 398 (50) 277 (29) 29 (44) 704 (39) <0.001
Province's towns without POU 252 (32) 323 (34) 25 (38) 600 (33)
Other Provinces 139 (18) 349 (37) 14 (21) 502 (28)
Total 789 (44) 949 (53) 68 (4) 1806 (100)
Diagnosis with pathology confirmation 758 (96) 916 (96) 66 (97) 1740 (96) 0.817
Hematological malignancies 382 (48) 479 (50) 40 (59) 901 (50) 0.220
POU of treatment
Only private 776 (98) 452 (48) 5 (7) 1233 (68) <0.001
Public to private 0 (0) 197 (21) 19 (28) 216 (12)
Private to public 1 (0) 3 (0) 0 (0) 4 (0)
Only public 1 (0) 282 (30) 44 (65) 327 (18)
POU outside the province 12 (2) 16 (2) 0 (0) 28 (2)
Total 790 (44) 950 (53) 68 (4) 1808 (100)
Diagnosis year
2009-10 218 (28) 226 (24) 55 (81) 499 (28) <0.001
2011-12 181 (23) 234 (25) 6 (9) 421 (23)
2013-14 209 (26) 254 (27) 7 (10) 470 (26)
2015-16 182 (23) 236 (25) 0 (0) 418 (23)
Total 790 (44) 950 (53) 68 (4) 1808 (100)

n ¼ 1808 if not otherwise stated.

*Uninsured.
 O. Ramirez et al. / Pediatric Hematology Oncology Journal 3 (2018) 79e87 83

Table 3
Causes of death in solid and hematological tumors by health insurance type
In solid tumors, treatment related deaths were similar in the public and private groups, but much higher in uninsured. In hematological tumors, as compared to the private
group, induction related death was three-times more frequent in public group and seven-times in the uninsured group. Treatment related deaths after induction were similar
for both private and public, but twice higher for the uninsured. Disease related deaths were twice higher in public and the uninsured group. Unknown causes of death were
almost evenly distributed among insurance status in solid tumors, with a higher frequency in the hematological tumors.

Solid tumors Health Insurance Category

Private Public *None Total

n (%) n (%) n (%) n (%)

Alive 262 (64) 275 (58) 10 (36) 547 (60)

Death 146 (36) 196 (42) 18 (64) 360 (40)
At diagnosis prior to cancer treatment 4 (1) 11 (2) 0 (0) 15 (2)
Treatment-related 9 (2) 6 (1) 14 (50) 29 (3)
Disease related 107 (26) 141 (30) 0 (0) 248 (28)
Second neoplasm 2 (0) 2 (0) 0 (0) 4 (0)
Other 0 (0) 0 (0) 4 (14) 4 (0)
Unknown cause 24 (6) 36 (8) 1 (4) 61 (7)
Total P ¼ 0.059 408 (45) 471 (52) 28 (3) 907 (100)
Hematological tumors
Alive 288 (75) 291 (61) 18 (45) 597 (66)
Death 94 (25) 188 (39) 22 (55) 304 (34)
At diagnosis prior to cancer treatment 7 (2) 8 (2) 0 (0) 15 (2)
Treatment-related during induction 9 (2) 30 (6) 6 (15) 45 (5)
Treatment related post-induction 22 (6) 34 (7) 6 (15) 62 (7)
Disease related 36 (9) 73 (15) 9 (23) 118 (13)
Second neoplasm 5 (1) 1 (0) 0 (0) 6 (1)
Other 1 (0) 0 (0) 0 (0) 1 (0)
Unknown cause 14 (4) 42 (9) 1 (3) 57 (6)
Total ¶P < 0.001 382 (42) 479 (53) 40 (4) 901 (100)

Likelihood-ratio chi2 statistical test for the frequency distribution.

*Uninsured.

public insurance was 20% (95% CI: 16, 24) and in uninsured patients private insurance, patients with public insurance, and uninsured
was 30% (95% CI: 16, 58). patients were 72% (95% CI: 63, 79), 46% (95% CI: 38, 52) and 38%
Five-year OS rates for all patients combined, children, and ad- (95% CI: 21 19, 57), respectively. Three-year OS rates for ALL in
olescents were 51% (95% CI: 48, 53), 52% (95% CI: 50, 55) and 41% children was 65% (95% CI: 60, 69); and for patients with private
(95% CI: 33, 48), respectively. Five-year OS for solid tumors was 47% insurance was 83%(95%CI: 76, 88) compared to patients with public
(95% CI: 42, 50) and for hematological tumors was 54% (95% CI: 50, insurance which was 52% (95% CI: 46, 59). Table 4 shows EFS results
58). Fig. 2 shows OS curves by health insurance type and by tumor by insurance type and tumor type. Mortality was higher in unin-
type. Three-year OS rates for children was 57% (95% CI: 55, 60); and sured patients (aHR 2.7; 95% CI: 1.9, 4.0), and in patients with
for patients with private insurance was 70% (95% CI: 66, 74) private insurance (aHR 1.6; 95% CI: 1.3, 1.9), compared to patients
compared to patients with public insurance, which was 49% (95% with private insurance, across the three major pathological groups
CI: 45, 53). Specifically, in ALL, the 5-year OS rates in patients with (acute leukemias, extracranial lymphoid and solid tumors and

Hematological malignancies Solid tumors

1

1
.1 .2 .3 .4 .5 .6 .7 .8 .9

.1 .2 .3 .4 .5 .6 .7 .8 .9
Survival probability

Survival probability
0

0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
Time since diagnosis (months) Time since diagnosis (months)
Number at risk Number at risk
Private 382 297 238 197 153 123 88 62 51 23 Public 408 295 226 179 134 99 67 42 24 10
Public 479 313 238 167 128 92 63 48 27 18 Private 471 274 183 130 91 67 47 27 11 1
Uninsured 40 22 18 14 12 8 8 7 4 2 Uninsured 28 13 6 4 3 2 1 1 1 1

Private Public Uninsured

Fig. 2. Overall survival graphs by tumor and health insurance type

This figure shows OS estimates for the cohort by health insurance and tumor type. In the hematological tumor group, the 5-year OS for private insurance was of 67.7% (95% CI: 61.8,
72.8), for public insurance was of 45.6% (95% CI: 40.2, 50.7), and for the uninsured group of 27.6% (95% CI: 14.5, 42.4). In the solid tumor group, the 5-year OS in patients with private
insurance patients with public insurance and uninsured patients 56.6% (95% CI: 50.8, 61.9), 39.1% (95% CI: 33.5, 44.6) and 14.1% (95% CI: 2.6, 34.8), respectively.
84 O. Ramirez et al. / Pediatric Hematology Oncology Journal 3 (2018) 79e87

Table 4
Five-year event free survival by tumor and health insurance category.

Classification (ICCC-3a) Five-year Event Free Survival

Health Insurance Category Total

Private Public

n % (95% CI) n % (95% CI) n % (95% CI)

I. Leukemias, myeloproliferative diseases, and myelodysplastic diseases 290 57 (50, 64) 384 36 (30, 42) 709 45 (41, 50)
II. Lymphomas and reticuloendothelial neoplasms 91 68 (56, 78) 95 64 (52, 73) 191 66 (58, 73)
III. CNS and miscellaneous intracranial and intraspinal neoplasms 142 42 (33, 51) 165 27 (17, 39) 316 35 (29, 42)
IV. Neuroblastoma and other peripheral nervous cell tumors 23 41 (19, 61) 18 23 (6, 46) 43 32 (16, 48)
V. Retinoblastoma 26 80 (59, 91) 45 61 (44, 75) 73 69 (56, 79)
VI. Renal tumors 34 74 (55, 86) 40 51 (33, 66) 76 62 (49, 73)
VII. Hepatic tumors 14 68 (34, 87) 11 58 (23, 82) 26 63 (39, 80)
VIII. Malignant bone tumors 57 42 (27, 57) 77 20 (10, 33) 136 30 (21, 40)
IX. Soft-tissue and other extraosseous sarcomas 42 46 (30, 61) 43 24 (11, 40) 89 36 (25, 47)
X. Germ cell tumors, trophoblastic tumors, and neoplasms of gonads 39 72 (53, 84) 38 52 (33, 68) 80 62 (49, 73)
XI. Other malignant epithelial neoplasms and malignant melanomas 25 87 (66, 96) 24 52 (27, 72) 51 72 (56, 83)
Total 783 57 (53, 61) 940 39 (35, 43) 1790 47 (45, 50)
a
ICCC-3: International classification of childhood cancer 3rd version. Total analysis time at risk ¼ 48067.8 months; Median time of follow-up in alive patients ¼ 30.6
months.

central nervous system tumors [CNS]) and after adjusting for the Table 5
confounders described in the methods section (see Table 5). Adjusted health insurance HR estimates for death from all causes.

Adolescents with CNS tumors had lower mortality (aHR 0.4; 95% Variables All-cause mortality n ¼ 1805
CI: 0.2, 0.8), whereas uninsured and public insurance patients with q
D HR (95% CI)
CNS tumors had increased mortality risk but with smaller aHR,
Health insurance
when compared with the estimates from the other two major
Private 253 1.0 ¥
histopathological types (Table 6). Residing outside Cali was strongly Public 447 1.6 (1.3,1.9)
associated with higher mortality risk in patients with CNS tumors Uninsured 46 2.7 (1.9,4.0)
(aHR 1.6; 95% CI: 1.3, 1.9). *Age group
Children 619 1.0
We found treatment center effects with worse outcomes for
Adolescents 127 1.3 (1.0,1.6)
patients who were treated in the public POU (aHR 1.6; 95% CI: 1.2, Sex
1.9). The mortality risk decreased by at least 20% (aHR 0.8; 95% CI: Female 311 1.0
0.6, 1.0) when patients were transferred from a public to a private Male 435 1.1 (0.9,1.2)
POU during treatment. We also found a 2-fold increase in mortality Afro-descendants
No 524 1.0
in patients with metastatic lymphoid and extracranial solid tumors
Yes 112 1.3 (1.1,1.6)
(aHR 2.3; 95% CI: 1.9, 4.0) (Table 6). Missing 110 1.2 (0.9,1.5)
Place of origin
4. Discussion Province's capital city with POU 271 1.0
Province's towns without POU 254 1.2 (1.0,1.4)
Other Provinces 221 1.3 (1.1,1.6)
In a comprehensive population-based analysis of childhood x
POU of treatment
cancer outcomes in a major city in Colombia, we found significant Only private 447 1.0
disparities in survival, primarily by insurance, despite universal Transfer from public to private 95 0.8 (0.6,1.0)
Only public 198 1.6 (1.2,1.9)
coverage for all children with cancer and access to uniform
Transfer to POU outside the province 6 0.6 (0.3,1.5)
chemotherapy protocols in Cali. Although we identified a treatment Period
center effect, this effect did not explain these disparities, as the aHR 2009-10 231 1.0
associated with health insurance type and treatment at public POU 2011-12 185 1.0 (0.8,1.2)
were independent. 2013-14 195 0.9 (0.8,1.2)
2015-16 135 1.0 (0.8,1.3)
Uninsured patients had inferior outcomes suggesting that uni-
Tumor classification
versal coverage did improve survival for those patients who were Hematological tumors 356 1.0
insured. However, universal coverage was insufficient to abolish VExtracranial solid tumors 235 1.0 (0.9,1.2)
survival disparities even among those with insurance, as we CNS tumors 155 1.4 (1.2,1.7)
observed a 3-fold higher risk of death in patients with public in- q D: Number of deaths.
surance compared to patients with private insurance. In adoles- ¥ Reference group.
cents and young adults in the U.S. with metastatic cancer and public *Age group: Children 0e14.9, adolescents 15e18.9 years of age.
xPOU: Pediatric oncology unit.
insurance, mortality was 50e70% higher compared to patients with
CNS: Central nervous system.
private insurance and after adjusting for sociodemographic factors
[27]. We found similar results in our cohort, where adolescents,
afro-descendants, patients who resided outside Cali, patients with population with private insurance. This could suggest barriers to
CNS tumors, and patients with metastatic disease had also higher timely diagnosis and referral to a treatment center in patients
mortality. younger than one year.
The lower proportion of patients younger than one year of age in In contrast to reports from HICs [28], we found that adolescents
our cohort among individuals with public insurance was an unex- were less frequently uninsured. Among the afro-descendant
pected finding, because the total number of infants in the popula- group, patients were more frequently uninsured or had public in-
tion covered by public insurance nationally was higher than in the surance compared to private insurance, consistent with insurance
 O. Ramirez et al. / Pediatric Hematology Oncology Journal 3 (2018) 79e87 85

Table 6
Adjusted health insurance HR estimates for death from all causes by type of tumor.

Variables All-cause mortality

¶
Acute leukemias Extracranial and lymphoid tumors CNS tumors

n ¼ 682 n ¼ 519 n ¼ 313

q
D HR (95% CI) D HR (95% CI) D HR (95% CI)

Health insurance
Private 83 1.0 ¥ 66 1.0 69 1.0
Public 195 2.0 (1.5, 2.8) 120 1.6 (1.1, 2.4) 75 1.3 (0.9, 1.9)
Uninsured 21 2.4 (1.4, 4.3) 8 6.4 (2.7, 15.1) 8 3.6 (1.5, 8.7)
Stage IV disease
No NA 101 1.0 NA
Yes 93 2.3 (1.7, 3.0)
*Age group
Children 246 1.0 166 1.0 142 1.0
Adolescents 53 1.9 (1.3, 2.9) 28 1.5 (1.0, 2.2) 10 0.4 (0.2, 0.8)
Sex
Female 122 1.0 86 1.0 61 1.0
Male 177 1.1 (0.9, 1.4) 101 1.1 (0.8, 1.4) 91 0.9 (0.7, 1.3)
Afro-descendants
No 218 1.0 155 1.0 107 1.0
Yes 48 1.5 (1.1, 2.1) 35 1.3 (0.9, 1.9) 13 0.7 (0.4, 1.2)
Missing 33 1.5 (0.9, 2.5) 4 2.0 (0.7, 5.3) 32 1.1 (0.7, 1.7)
Place of origin
Province's capital city with POU 105 1.0 68 1.0 64 1.0
Province's towns without POU 104 1.0 (0.7, 1.3) 63 1.0 (0.7, 1.5) 47 1.2 (0.8, 1.8)
Other Provinces 90 1.1 (0.8, 1.5) 63 1.3 (0.9, 1.9) 41 1.5 (1.0, 2.2)
§
POU of treatment
Only private 158 1.0 122 1.0 116 1.0
Transfer from public to private 64 0.8 (0.6, 1.5) 21 0.6 (0.3, 0.9) 2 0.3 (0.1, 1.2)
Only public 76 1.5 (1.1, 2.2) 48 1.7 (1.2, 2.7) 33 0.9 (0.6, 1.5)
Transfer to POU outside the province 1 0.3 (0.1, 2.0) 3 1.3 (0.4, 4.2) 1 0.4 (0.1, 3.1)
Period
2009-10 90 1.0 205 1.0 45 1.0
2011-12 78 1.3 (0.9, 1.7) 159 0.9 (0.6, 1.5) 36 0.7 (0.5, 1.2)
2013-14 83 1.0 (0.7, 1.4) 171 0.9 (0.6, 1.5) 43 0.7 (0.4, 1.1)
2015-16 48 0.9 (0.6, 1.3) 110 1.6 (1.0, 2.5) 28 0.7 (0.4, 1.2)

Q D: Number of deaths.
¥ Reference group.
* Age group: Children 0e14.9, adolescents 15e18.9 years of age.
x POU: Pediatric oncology unit.
CNS: Central nervous system.

disparities associated with race reported in HIC [27e30]. The pro- patients with public insurance compared to patients with private
portion of uninsured patients decreased over time during the study insurance. Treatment abandonment and refusal are the most
period with no reports of uninsured patients after 2015. extreme form of lack of treatment adherence. We did not have
We found a 2-fold increase of metastatic disease at diagnosis in adherence measurements in our study, however, the role of cu-
patients with public insurance and extracranial tumors, with a mulative delays in chemotherapy administration could explain
strong association in patients with malignant bone and soft tissue survival disparities, as previously described [35]. Treatment
tumors. Stage at diagnosis is one of the determinant factors for adherence and abandonment are complex social phenomena [36],
survival disparities in adults with cancer among different SES which involve family income, social support networks, cultural
groups [26]. Stage at diagnosis in extracranial solid tumors can be beliefs and family values, and the ability to navigate complex
used as a surrogate measurement of timely access to high quality healthcare systems, such as the one in Colombia.
care because clinical outcomes are highly dependent on disease Event-free survival disparities were consistent among all ICCC-3
extension in this type of tumors [31]. Advanced stage at diagnosis groups of tumors, with the exception of lymphomas, CNS and
implies long delays between disease suspicion by the family to epithelial tumors. Lymphomas (particularly Hodgkin lymphoma)
diagnostic confirmation of cancer. are highly curable with short treatments, including those with
In hematological tumors, treatment-related mortality during advanced disease. Therefore, in this subgroup, the impact of time to
induction was 2.5-fold higher in patients with public insurance diagnosis, treatment-related mortality, and treatment abandon-
compared to patients with private insurance. Moreover, treatment- ment were lower and, hence, patients with public insurance had
related mortality post-induction in patients with ALL remained better survival rates overall. In contrast, EFS in patients with CNS
significantly higher in patients with public insurance. Noteworthy, tumors was lower in patients with both public and private insur-
both treatment-related mortality during induction and post- ance. This is explained by the complexity of treatment for these
induction were higher in patients with private and public insur- tumors and its reliance on very sophisticated and complex care,
ance when analyzed combined, compared to patients in HIC which is limited in LMIC. We also found a higher OS in adolescent
[32,33]. Many of these deaths could be attributed to the limited patients with CNS tumors compared to younger patients. This dif-
healthcare system infrastructure to adequately support children ference could be attributed to the particular biology of CNS tumors
with neutropenic fever, sepsis and septic shock [33,34]. in infants, the avoidance of radiotherapy as a treatment strategy in
We found a 6-fold increase in treatment abandonment in this age group, and the complexity of the care in younger patients.
86 O. Ramirez et al. / Pediatric Hematology Oncology Journal 3 (2018) 79e87

Five-year EFS for thyroid cancer -the most frequent epithelial tumor pediatric cancer registries in LMICs, to guide local health policy
in children and adolescents and highly curable-was lower in our decisions aimed at improving outcomes in resource-constrained
patients with public insurance than the one reported in the liter- settings and help narrow the survival gap.
ature [37]. This finding could be associated to the limited avail- We hope that this data-driven report on survival disparities can
ability of radioactive iodine therapy in Cali. guide policy-makers and healthcare insurance leaders in our
In patients with hematological tumors, high rates of treatment- country and other resource-strained settings in identifying gaps in
related mortality, particularly during induction, and treatment care by insurance type and plan infrastructure and system changes
abandonment explain most of the survival disparities found among to achieve equitable care and outcomes for all.
patients with different insurance types. In extracranial solid tu-
mors, advanced stages at diagnosis and treatment abandonment Acknowledgements
seem to play a more prominent role in survival disparities than
treatment-related mortality. Supported by the My Child Matters program of Sanofi-Espoir
At the regional level, the Argentinian hospital-based cancer Foundation (O.R., A.Z and A.G.) and the POHEMA Foundation (O.R.
registry (ROHA) reported a three-year OS for children with all and L.E.B.); and the American Lebanese Syrian Associated Charities
cancer types of 62% (95% CI: 61, 63), in the 2000e2007 cohort [38]. (R.C.R.). We thank the Sanofi-Espoir Foundation leadership for their
The 57% three-year OS found on our population is comparable to support and for facilitating access to international experts who
the reported by ROHA, taking into account that Argentina is a HIC mentored and guided this project through the My Child Matters
[39]. We report a 70% three-year OS in patients with private in- program. We thank Cali's population-based cancer registry staff Luz
surance that is very similar to the OS reported in the capital city of Estela García and Paola Collazos and collaborators of the VIG-
Buenos Aires. However, the 49% 3-year OS in patients with ICANCER Group and POHEMA Foundation: Margarita Quintero,
public insurance is below the lowest OS in Argentina of 55% in the Carlos Portilla, Viviana Lotero, Diego Medina, Ximena Castro, Luz
north-west region. For ALL, the five-year OS of 72% in patients with

Angela Urcuqui, Mo
nica Lotero and Dilia Escobar.
private insurance is comparable to Argentina's (76%), higher than
the one reported in Brazil (66%), Chile (64%), Mexico (53%) and
References
Ecuador (50%), but lower compared to the one reported in Costa
Rica (80%) [4]. [1] Gatta G, Botta L, Rossi S, Aareleid T, Bielska-Lasota M, Clavel J, et al. Childhood
Our study has some limitations and the generalizability of the cancer survival in Europe 1999e2007: results of EUROCARE-5ea population-
based study. Lancet Oncol 2014;15:35e47.
results may be limited given that the population studied was cir-
[2] Ward E, DeSantis C, Robbins A, Kohler B, Jemal A. Childhood and adolescent
cumscribed to one city. We were not able to directly assess the cancer statistics. CA Cancer J Clin 2014;64:83e103. 2014.
effects of SES on clinical outcomes, although health insurance type [3] Howard SC, Zaidi A, Cao X, Weil O, Bey P, Patte C, et al. The My Child Matters
is inevitable linked to SES [15,16,25,28,40], and therefore could be a programme: effect of publiceprivate partnerships on paediatric cancer care in
low-income and middle-income countries. Lancet Oncol 2018;19:252e66.
mediator between the SES effect and survival [25]. Although, SES is [4] Allemani C, Matsuda T, Di Carlo V, Harewood R, Matz M, Niksi
c M, et al. Global
a multi-dimensional and complex construct [25], by including afro- surveillance of trends in cancer survival 2000e14 (CONCORD-3): analysis of
descendant ancestry and place of residence in the analyses, we individual records for 37 513 025 patients diagnosed with one of 18 cancers
from 322 population-based registries in 71 countries. Lancet 2018;391:
could partially adjust the health insurance effect estimate for SES. 1023e75.
National statistics report that afro-descendants in Colombia have [5] Henderson TO, Bhatia S, Pinto N, London WB, McGrady P, Crotty C, et al. Racial
low education attainment, are more frequently unemployed and a and ethnic disparities in risk and survival in children with neuroblastoma: a
Children's Oncology Group study. J Clin Oncol 2011;29:76e82.
great majority live under the poverty line [16]. [6] Wang L, Bhatia S, Gomez SL, Yasui Y. Differential inequality trends over time
Place of residence, reflecting the distance to the treatment in survival among U.S. children with acute lymphoblastic leukemia by race/
center and travel time, was associated with an increased risk of ethnicity, age at diagnosis, and sex. Cancer Epidemiol Biomark Prev 2015;24:
1781e8.
death, which was even higher if the patient resided in distant
[7] Bhatia S. Disparities in cancer outcomes: lessons learned from children with
provinces. Finally, travel time or distance to treatment center was cancer. Pediatr Blood Canc 2011;56:994e1002.
not available, thus we used place of residence to partially adjust for [8] Aristizabal P, Fuller S, Rivera-Gomez R, Ornelas M, Nuno L, Rodriguez-
Galindo C, et al. Addressing regional disparities in pediatric oncology: results
distance to treatment center [41].
of a collaborative initiative across the Mexican-North American border.
In 2014, 97% of Colombian citizens were covered by the national Pediatr Blood Canc 2016;64. https://doi.org/10.1002/pbc.26387.
compulsory health insurance system [16]. Although this universal [9] Rodriguez-Galindo C, Friedrich P, Morrissey L, Frazier L. Global challenges in
coverage constitutes an enormous public health achievement and pediatric oncology. Curr Opin Pediatr 2013;25:3e15.
[10] World health assembly resolution 58.33. Geneva: WHO; 2005. World Health
has increase access to basic care, significant survival disparities Organization Web site] Available at: http://www.who.int/health_financing/
remain. This inequity in a setting with a universal healthcare sys- documents/cov-wharesolution5833/en/. [Accessed 9 February 2018].
tem is, unfortunately, not only common in LMIC, such as Colombia [11] Rodney AM, Hill PS. Achieving equity within universal health coverage: a
narrative review of progress and resources for measuring success. Int J Equity
but also in HICs with similar health insurance systems Health 2014;13:72e80.
[25e27,42e46]. A major challenge for national health systems is [12] Hanratty B, Zhang T, Whitehead M. How close have universal health systems
not only to ensure that all populations have adequate access to come to achieving equity in use of curative services? A systematic review. Int J
Health Serv 2007;37:89e109.
health coverage and services [47,48], but also to create truly high- [13] De Groote T, De Paepe P, Unger JP. Colombia: in vivo test of health sector
quality care through the continuum of care [48e50]. To achieve privatization in the developing world. Int J Health Serv 2005;35:125e41.
this purpose, itis essential to identify disparities in outcomes at the [14] Garcia-Subirats I, Vargas I, Mogollo
n-Pe
rez AS, De Paepe P, da Silva MR,
Unger JP, et al. Barriers in access to healthcare in countries with different
population-based level and prioritize resources guided by data health systems. A cross-sectional study in municipalities of central Colombia
driven-evidence; develop healthcare infrastructure and train hu- and north-eastern Brazil. Soc Sci Med 2014;106:204e13.
man capital; create national cancer control plans; implement or [15] Plaza B, Barona AB, Hearst N. Managed competition for the poor or poorly
managed competition? Lessons from the Colombian health reform experi-
improve established cancer registries; support the creation of
ence. Health Policy Plan 2001;16:44e51.
reference, comprehensive cancer centers; define resource- [16] De La Cruz R, Andri
an LG, Loterzpil M. Social mobility and the consolidation of
stratified evidence-based diagnosis, and treatment of common the middle class. Colombia: towards a high-income country with high social
cancers; and guarantee access to essential cancer medicines and to mobility, vols. 93e121. Inter-American Development Bank; 2015, ISBN 978-1-
59782-265-7. Available at: https://publications.iadb.org/handle/11319/7435.
palliative care [48,50]. [Accessed 12 July 2018].
Our findings underscore the importance of establishing [17] Bravo LE, García LS, Collazos P, Aristizabal P. Descriptive epidemiology of
 O. Ramirez et al. / Pediatric Hematology Oncology Journal 3 (2018) 79e87 87

childhood cancer in Cali, Colombia 1977e2011. Colomb Me
d 2013;44: infectious deaths on UKALL2003. Blood 2014;124:1056e61.
155e64. [35] Suarez A, Pin~ a M, Nichols-Vinueza DX, Lopera J, Rengifo L, Mesa M, et al.
[18] Ramirez O, Aristizabal P, Alia Z, Ribeiro RC, Bravo LE, VIGICANCER Working A strategy to improve treatment-related -mortality and abandonment of
Group. Implementing a childhood cancer outcomes surveillance system therapy for childhood ALL in a developing country reveals the impact of
within a population-based cancer registry. Journal of Global Oncology 2018;4: treatment delays. Pediatr Blood Canc 2015;62:1395e402.
1e11. [36] Ospina-Romero M, Portilla CA, Bravo LE, Ramirez O, VIGICANCER working
[19] Montenegro T, Acevedo F, Bernal O. Colombia case study: the subsidized group. Caregivers' self-reported absence of social support networks is related
regime of Colombia's national health insurance system. In: Universal Health to treatment abandonment in children with cancer. Pediatr Blood Canc
Coverage (UNICO) studies series ; no. 15. Washington, DC: World Bank Group; 2016;63:825e31.
2013 [The World Bank Web site], http://documents.worldbank.org/curated/ [37] Trama A, Botta L, Foschi R, Ferrari A, Stiller C, Desandes E, et al. Survival of
en/727721468239997995/Colombia-case-study-the-subsidized-regime-of- European adolescents and young adults diagnosed with cancer in 2000e07:
Colombias-national-health-insurance-system. [Accessed 9 February 2018]. population-based data from EUROCARE-5. Lancet Oncol 2016;17:896e906.
[20] Burton A. The My Child Matters awards: new funding, new countries, new [38] Moreno F, Veronica D, Orellana L, ROHA network. Childhood ca
ncer in
hope. Lancet Oncol 2009;10:216e7. Argentina: survival 2000-2007. Cancer Epidemiol 2015;39:505e10.
[21] World Health Organization. In: International classification of diseases for [39] World Bank. World bank country and lending groups. Available at: https://
oncology, vol. 3. Geneva, Switzerland: World Health Organization; 2013. ICD- datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-
O-3) 1st revision. country-and-lending-groups. [Accessed 17 December 2018].
[22] WHO Classification of tumours of haematopoietic and lymphoid tissues. In: [40] Adler NE, Newman K. Socioeconomic disparities in health: pathways and
Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Stein H, et al., editors. policies. Health Aff 2002;21:60e76.
IARC WHO classification of tumours, vol. 2. Geneva, Switzerland: World [41] Metzger ML, Howard SC, Fu LC, Pen ~ a A, Stefan R, Hancock ML, et al. Outcome
Health Organization; 2008. 4. of childhood acute lymphoblastic leukaemia in resource-poor countries.
[23] Steliarova-Foucher E, Stiller C, Lacour B, Kaatsch P. International classification Lancet 2003;362:706e8.
of childhood cancer, third edition. Cancer 2005;103:1457e67. [42] Sommers BD, McMurtry CL, Blendon RJ, Benson JM, Sayde J. Beyond health
[24] Mostert S, Arora RS, Arreola M, Bagai P, Friedrich P, Gupta S, et al. Aban- insurance: remaining disparities in U.S. health care in the post-ACA era.
donment of treatment for childhood cancer: position statement of a SIOP Milbank Quarterly. Forthcoming 2017:1e45. Available at: http://nrs.harvard.
PODC Working Group. Lancet Oncol 2011;12:719e20. edu/urn-3:HUL.InstRepos:29695273. [Accessed 10 October 2018].
[25] Gupta S, Wiletjo M, Pole JD, Guttmann A, Sung L. Low socioeconomic status is [43] Bambra C, Garthwaite K, Hunter D. All things being equal: does it matter for
associated with worse survival in children with cancer: a systematic review. equity how you organize and pay for health care? Rev. Int. Evid. Int. J. Health
PLoS One 2014;9:e89482. Serv. 2014;44:457e77.
[26] Woods LM, Rachet B B, Coleman MP. Origins of socio-economic inequalities in [44] Serapioni M. Economic crisis and inequalities in health systems in the coun-
cancer survival: a review. Ann Oncol 2006;17:5e19. tries of Southern Europe. Cad Saúde Pública 2017;33:e00170116.
[27] DeRouen MC, Parsons HM, Kent EE, Pollock BH, Keegan THM. Sociodemo- [45] Nipp RD, Shui AM, Perez GK, Kirchhoff AC, Peppercorn JM, Moy B, et al. Pat-
graphic disparities in survival for adolescents and young adults with cancer terns in health care access and affordability among cancer survivors during
differ by health insurance status. Cancer Causes Control 2017;28:841e51. implementation of the Affordable Care Act. JAMA Oncol 2018;4:791e7.
[28] Rosenberg AR, Kroon L, Chen L, Li CI, Jones B. Insurance status and risk of [46] Feller A, Schmidlin K, Bordoni A, Bouchardy C, Bulliard JL, Camey B, et al.
cancer mortality among adolescents and young adults. Cancer 2015;121: Socioeconomic and demographic disparities in breast cancer stage at pre-
1279e86. sentation and survival: a Swiss population-based study. Int J Cancer
[29] Aizer AA, Falit B, Mendu ML, Chen M, Choueiri TK, Hoffman KE, et al. Cancer- 2017;141:1529e39.
specific outcomes among young adults without health insurance. J Clin Oncol [47] Asante A, Price J, Hayen A, Jan S, Wiseman V. Equity in health care financing in
2014;32:2025e30. low- and middle- income countries: a systematic review of evidence from
[30] Alvarez E, Keegan T, Johnston EE, Haile R, Sanders L, Saynina O, et al. studies using benefit and financing incidence analyses. PLoS One 2016;11:
Adolescent and young adult oncology patients: disparities in access to e0152866.
specialized cancer centers. Cancer 2017;123:2516e23. [48] Patel V, Parikh R, Nandraj S, Balasubramaniam P, Narayan K, Paul VK, et al.
[31] Dang-Tang T, Franco EL. Diagnosis delays in childhood cancer: a review. Assuring health coverage for all in India. Lancet 2015;386:2422e35.
Cancer 2007;110:703e13. [49] Knaul FM, Bhadelia A, Atun R, Frenk J. Achieving effective universal health
[32] Slats AM, Egeler RM, van der Does-van den Berg A, Korbijin C, Ha €hlen K, coverage and diagonal approaches to care for chronic illnesses. Health Aff
Kamps WA, et al. Causes of death e other than progressive leukemia e in (Millwood) 2015;34:1514e22.
childhood acute lymphoblastic (ALL) and myeloid leukemia (AML): the Dutch [50] El Saghir NS, Farhat RA, Charara RN, Khoury KE. Enhancing cancer care in
Childhood Oncology Group experience. Leukemia 2005;19:537e44. areas of limited resources: our next steps. Future Oncol 2014;10:1953e65.
[33] Prucker C, Attarbaschi A, Peters C, Dworzak MN, Po €tschger U, Urban C, et al. [51] National and regional demographic, social and economical statistics. Colom-
Induction death and treatment-related mortality in first remission of children bian National Statistics Bureau (DANE): Available at: https://www.dane.gov.
with acute lymphoblastic leukemia: a population-based analysis of the Aus- co/index.php/estadisticas-por-tema/demografia-y-poblacion. Accessed 11 02,
trian Berlin-Frankfurt-Münster study group. Leukemia 2009;23:1264e9. 2018.
[34] O'Connor D, Bate J, Wade R, Clack R, Dhir S, Hough R, et al. Infection-related [52] Cali's demography, social and economical statistics. Cali's planning office.
mortality in children with acute lymphoblastic leukemia: an analysis of Available at: http://planeacion.cali.gov.co/sis/indexAccessed 11 02, 2018.