Abstract

INTRODUCTION

Retinoblastoma (RB), the most common primary intraocular malignancy affecting infants and children, represents about 4% of paediatric malignancies.^{1 2} Worldwide, the incidence of RB has been estimated to be 1:14 000–18 000 live births (~8000 children globally each year), with mortality of >3000 children annually.^{2 3} Rates of mortality are disproportionately higher in low-income and middle-income countries (LMICs; mortality: 40%–70%) compared with high-income countries (HICs; mortality: 3%–5%).^{3 4} The prognosis of RB has improved in HICs (>95% of disease-free survival rates) due to factors such as increased specialisation centres, improved screening and awareness, and availability of new treatment regimens.^5–7 The prognosis in LMICs remains guarded due to delay in diagnosis and treatment or abandonment of care attributed to various factors, such as socioeconomic status (SES) and healthcare access.^{4 8 9}

Of the paediatric cancers occurring worldwide, it is estimated that only 20%–30% are diagnosed and treated, with most of these being treated in HICs (>80%).^10–13 Rates of care abandonment in paediatric cancer are highest in LMICs, ranging up to 60% in some studies.^{14 15} Abandonment of paediatric cancer care in LMICs is correlated with country income level, parental educational status, travel times, prognosis of the disease, and care affordability and accessibility.^{14 16–21} In RB specifically, the data on abandonment are varied, with single-centre reports showing a range in care abandonment of 38% (35 of 91) in Tanzania, 22%–35% in Central America and 50% in India.^22–25 Studies from India have investigated risk factors for abandonment and have shown that abandonment is increased in rural children, financial concerns, unwillingness to enucleate, female sex, bilateral disease and difficulty in attending outpatient appointments.^{20 26 27} The prevalence and risk factors associated with abandonment of care in RB have not been studied in a multinational cohort. The purpose of this study is to investigate abandonment of care in RB across multiple countries, continents and healthcare systems in order to identify factors associated with care abandonment.

MATERIALS AND METHODS

Details of the methodology for data collection in this prospective observational study have been described in detail in Kaliki et al.²⁸ Briefly, clinical and demographic information was gathered prospectively for all patients with newly diagnosed RB who presented to 11 international centres within 10 countries (Bangladesh, China, Ethiopia, France, India, Pakistan, Peru, Russia, UK, USA) during the calendar year 2019. Centres were selected from all continents and income ranges and they represent centres where prospective data could be collected over 1 year.

The outcome of interest was abandonment of care. Centres were asked specifically if each child’s care was abandoned. Detailed information about the reasons for abandonment or the type of abandonment was extracted from the questionnaire’s free-text comments. Patient care was considered ‘abandoned’ if the child was lost to follow-up and did not choose to obtain care from another known provider. Variables considered as potential risk factors for abandonment were included in a statistical model. These covariates included age, country, laterality, vision at presentation, clinical tumours staging (American Joint Committee on Cancer Eighth Edition),²⁹ presence of lymph node disease, presence of metastases, distance to treatment centre, sex, mother’s age, birth order, lag time between symptoms and presentation, number of physicians seen prior to arriving at the treatment centre, and enucleation as the primary treatment (labelled ‘Enucleation’).

RESULTS

Data from 692 patients from 10 countries and 11 treatment centres were included in the analysis. Table 1 shows the demographic characteristics and features of these patients stratified by the 10 countries. Of the patients, 75 abandoned care, or 11% of this cohort. The distribution of each variable, stratified by abandonment, is reported in table 2. Three countries reported care abandonment of over 10%: Pakistan, Bangladesh and Peru. Four countries reported no care abandonment: Russia, UK, France and USA.

Table 1

Distribution of demographics and features of 692 patients by country

	Patients, n (%)	Mean age at presentation (months)	Sex (male:female)	Care abandoned, n (%)	American Joint Committee on Cancer Eighth Edition tumour staging (worse eye)
					T1	T2	T3	T4
Total	692	24	369:323	75 (11)
Country
Bangladesh	136 (20)	20	71:65	42 (31)	4	21	100	11
China	166 (24)	22	82:84	5 (3)	6	90	60	10
Ethiopia	74 (11)	31	41:33	5 (7)	5	25	24	20
France	49 (7)	23	27:22	0 (0)	6	24	18	0
India	128 (18)	27	68:59	5 (4)	5	58	49	16
Pakistan	30 (4)	26	15:15	10 (33)	0	16	8	6
Peru	46 (7)	25	29:17	8 (17)	1	32	7	5
Russia	42 (6)	21	24:18	0 (0)	1	22	9	10
UK	14 (2)	18	9:5	0 (0)	1	11	2	0
USA	7 (1)	12	3:4	0 (0)	1	6	0	0

Open in a separate window

Table 2

Distribution of each covariate stratified by care abandonment

	Care not abandoned (n=617), n (%)	Care abandoned (n=75), n (%)	Overall (n=692), n (%)
Mean age at presentation (months)	24 (19)	24 (17)	24 (29)
Country
Bangladesh	94 (69)	42 (31)	136 (20)
China	161 (97)	5 (3)	166 (24)
Ethiopia	69 (93)	5 (7)	74 (11)
France	49 (100)	0 (0)	49 (7)
India	123 (96)	5 (4)	128 (18)
Pakistan	20 (67)	10 (33)	30 (4)
Peru	38 (83)	8 (17)	46 (7)
Russia	42 (100)	0 (0)	42 (6)
UK	14 (100)	0 (0)	14 (2)
USA	7 (100)	0 (0)	7 (1)
Continent
Europe	105 (100)	0 (0)	105 (15)
Asia	398 (87)	62 (13)	460 (66)
Africa	69 (93)	5 (7)	74 (11)
Latin America	38 (83)	8 (17)	46 (7)
North America	7 (100)	0 (0)	7 (1)
Country income
Low	69 (93)	5 (7)	74 (11)
Low-middle	237 (81)	57 (19)	294 (42)
Upper-middle	241 (95)	13 (5)	254 (37)
High	70 (100)	0 (0)	70 (10)
Distance to retinoblastoma centre (kilometers)	623 (988)	309 (435)	589 (59)
Sex
Male	334 (91)	35 (9)	369 (53)
Female	282 (88)	40 (12)	322 (47)
Mother’s age at birth (years)	27.7 (5.7)	25.9 (5.2)	27.5 (5.7)
Birth order	2.0 (1.2)	2.0 (1.3)	2.0 (1.2)
Number of physicians	1.4 (0.7)	1.6 (0.9)	1.4 (0.8)
Lag time (days)	145 (193)	191 (222)	150 (197)
Laterality
Right	200 (92)	18 (8)	218 (32)
Left	232 (86)	39 (14)	271 (39)
Both	184 (91)	18 (9)	202 (29)
Vision at presentation (number of eyes that can fix and follow)
Both	47 (98)	1 (2)	48 (11)
One	257 (83)	53 (17)	310 (74)
None	50 (82)	11 (18)	61 (15)
American Joint Committee on Cancer Eighth Edition staging (worse eye)
T1	28 (100)	0 (0)	28 (4)
T2	295 (97)	10 (3)	305 (44)
T3	227 (82)	50 (18)	277 (40)
T4	64 (82)	14 (18)	78 (11)
Lymph node disease
Not examined	277 (96)	12 (4)	289 (44)
No	289 (83)	58 (17)	347 (53)
Yes	16 (80)	4 (20)	20 (3)
Metastasis
No	581 (90)	61 (10)	642 (94)
Yes	29 (76)	9 (24)	38 (6)
Enucleation
No	328 (91)	31 (7)	359 (55)
Yes	285 (90)	31 (10)	316 (45)

Open in a separate window

The results of the multivariate logistic regression of complete cases are presented in table 3. The multivariate analysis identified four covariates that were independently associated with care abandonment: country (p<0.001), advanced clinical staging (OR 4.22, 95% CI 1.99 to 9.60, p<0.001), no metastases at presentation (OR 0.17, 95% CI 0.03 to 0.89, p=0.047; patients with metastasis were less likely to have care abandoned) and female sex (OR 2.39, 95% CI 1.21 to 4.84, p=0.013). On average, increasing clinical stage was associated with increasing odds of care abandonment (OR 4.22, 95% CI 1.99 to 9.60, p<0.001) in the group with higher clinical stage. When comparing a group of female patients with another group of male patients who were otherwise the same, the odds of care abandonment also increased (OR 2.39, 95% CI 1.21 to 4.84, p=0.013), higher in female patients. Notably, enucleation as the primary treatment (OR 0.59, 95% CI 0.26 to 1.33, p=0.206) was not statistically significant in its association with care abandonment.

Table 3

Multivariate logistic regression of associations between care abandonment and all covariates except continent, country income and vision at presentation for complete cases

Variable	OR (95% CI)	P value
Age at presentation (months)	1.00 (0.99 to 1.00)	0.331
Country*		<0.001
Bangladesh	18.1 (0.69 to 238)
China†	0.00 (−)
Ethiopia†	0.00 (−)
France†	0.00 (−)
Pakistan	45.5 (1.45 to 828)
Peru	9.23 (0.28 to 157)
Russia†	0.00 (−)
UK†	0.00 (−)
Distance (log-transformed)	0.94 (0.69 to 1.27)	0.666
Sex
Female	2.39 (1.21 to 4.84)	0.013
Mother’s age at birth (years)	0.99 (0.91 to 1.07)	0.755
Birth order	1.23 (0.83 to 1.87)	0.309
Number of physicians	0.91 (0.58 to 1.40)	0.666
Lag time (log-transformed)	0.87 (0.65 to 1.16)	0.344
Laterality		0.089
Left eye	2.28 (1.01 to 5.35)
Both eyes	1.09 (0.41 to 2.92)
Clinical staging	4.22 (1.99 to 9.60)	<0.001
Lymph node disease		0.823
No	0.50 (0.05 to 11.2)
Yes	0.36 (0.02 to 13.1)
Metastasis
Yes	0.17 (0.03 to 0.89)	0.047
Enucleation
Yes	0.59 (0.26 to 1.33)	0.206

Open in a separate window

Bolded values: p values <0.05 that are considered statistically significant

^*India is chosen as the reference as the rate of care abandonment ranks in the middle; there are no results for the USA as all patients have missing age, hence excluded from the multivariate regression.

^†All patients have retained their care in this country. Point estimate of the OR is 0. Wald-type CIs have length infinity and are not meaningful.

DISCUSSION

Studies from various countries have identified different factors to be statistically significant in effecting RB care abandonment, including distance, SES, patient sex, parental educational status and apprehension of enucleation.^{10 14 21 24 27 30} Multivariate modelling of our data identified that country, female sex and advanced clinical stage were positively associated with abandonment, whereas metastasis was negatively associated with abandonment.

It is widely reported in oncology literature that country is significantly associated with higher abandonment rates, which is supported by our composite international data. Abandonment of care or treatment refusal is often not analysed in HIC data because it is primarily seen in LMICs, usually attributed to resource constraints. Similarly, our study shows abandonment of care in every LMIC in the study, while none of the HICs showed abandonment. Similarly, Kaliki et al.²⁸ found country income level significantly impacts lag time for diagnosis of RB in this data set. Financial barriers have long remained astute in their role in impacting health outcomes. Nuances of these results should be further explored, particularly in relation to family income, costs of treatments and subsidies provided by the government or private entities for care provision and treatment. This information was not readily available to analyse in our data set but may be a potential confounder in our results. These data cannot discern a difference in abandonment of care based on overall cost of treatment, although this would be an important question in the future.

Advanced clinical stage at the time of presentation is associated with higher rates of abandonment. Of the patients who died during this study’s short follow-up period, nearly 50% of them had abandoned care, indicating increased mortality in this subsection. Assuming an equal risk between each stage, the OR from each successive clinical stage was 4.22. Conversely, the presence of metastasis at diagnosis was found to be a protective factor for care abandonment. The variables clinical tumour stage and metastasis are correlated, but the data suggest they are unique. The reason for this finding is unknown and more investigation is required.

Female sex is associated with higher rates of care non-compliance. Previous studies have come to various conclusions regarding sex and care abandonment. They range from finding child cancer care abandonment to be higher in LMICs, overall, with no sex bias,³¹ to finding female sex plays a minor role in childhood cancer treatment abandonment, but more so in LMICs than in HICs.³² Similarly, previous studies on RB have shown variable results in the significance of association between compliance and female sex.^{20 27 33}

Enucleation was not related to abandonment. It has long been known that apprehension for enucleation is a risk factor for abandonment in many countries. Many centres have altered their treatment philosophy to address this issue. A large referral centre in Uganda recently reported that its programme of recommending upfront chemotherapy, even when a globe is not salvageable, was found to reduce the rate of care abandonment and increase the likelihood that after a few initial cycles of chemotherapy, families would be more willing to accept enucleation.³⁴ This programme also included other features such as minimal hospital stay, provision of transportation and food costs, and benefits of peer support that may also be compounding factors that affect decisions for care retention.³⁴ Likely, centres enrolled in our study have adapted their treatment philosophies in similar ways, which introduces a statistical bias when analysing patients in whom enucleation was the primary treatment.

Limitations of this study include its limited sample size. While this is the largest multinational cohort that has addressed risk factors for care abandonment in RB, the sample size in some countries is relatively small, with an unequal distribution of patients among the countries. Each RB centre was chosen based on its ability to provide 1 year of detailed prospective data and opted into the study, introducing another level of bias in our results, especially with regard to ensuring appropriate comparisons by income brackets. Nevertheless, this is the most extensive study to date, with all income levels and most continents with statistically significant results, making it more widely applicable than other single-centred results. These issues are important to consider in future studies. Our multivariate analysis excluded vision at presentation because over 40% (263) of the data were missing. While it is possible that vision may be an additional predictor of care abandonment, studies have shown that vision is highly correlated with disease stage,³⁵ and this study has robust data on patient disease stage.

CONCLUSION

Findings of the present study show that the risk factors for abandonment of care in RB include the patient’s country of residence, advanced disease stage and female sex. The data suggest that international differences are more compelling and involved in RB abandonment than other factors. Importantly, enucleation as the primary treatment was not categorically associated with higher rates of abandonment in this study, but this may be due to current practice patterns. RB centres should be aware that advanced disease, female sex and lack of metastasis in children with advanced RB might be factors leading to subsequent care abandonment within their centres.

Funding

Footnotes

REFERENCES