Original Research

Two-year overall survival of children with cancer between 2015 and 2021 in Zimbabwe

Laura P. Valerio Sibanda, Anel van Zyl, Tonya M. Esterhuizen, Daniel Mckenzie, Loyce Hlatywayo, Patience Kuona, Vongai Dondo, Mariana Kruger

Received: 04 Aug. 2025; Accepted: 23 Oct. 2025; Published: 17 Dec. 2025

Copyright: © 2025. The Authors. Licensee: AOSIS.
This work is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license (https://creativecommons.org/licenses/by/4.0/).

Abstract

Background: Childhood cancer survival in low- and middle-income countries (LMICs) is low, and data are limited.

Aim: We investigated 1-year and 2-year overall survival (OS) of children diagnosed with cancer in Zimbabwe between 2015 and 2021.

Setting: Retrospective childhood cancer data from 2015 to 2022 were collected from KidzCan’s database.

Methods: Demographic data, cancer diagnosis and performance status were collected to determine OS 2 years post-cancer diagnosis. Data were analysed using Kaplan-Meier survival curves and Cox regression analysis.

Results: A total of 514 from 1090 childhood cancer cases were analysed (576 cases were excluded because of missing data). The mean age at diagnosis was 5.0 years (standard deviation 3.8), with a male-to-female ratio of 1:1.25. One-year OS was 38% (mean 7.5 months; 95% confidence interval [CI] [7.1–8.0]) and 2-year OS was 17.3% (mean 10 months; 95% CI [9.7–11.4]). Male cases (21.9%) had a better 2-year OS compared to female cases (10%) (p = 0.012). Lymphoma and retinoblastoma had the best 2-year OS (29% and 28%, respectively) (p = 0.002). Children active at diagnosis (28%) had an improved 2-year OS than those who were frail (11.5%) and bedridden (13%) (p < 0.001).

Conclusion: Sex, cancer diagnosis and performance status were associated with OS. Long-term follow-up is essential to determine long-term outcomes for all children treated with cancer. This study highlights the vital role of civil society organisations in supporting the establishment of a local childhood cancer registry.

Contribution: This study will significantly contribute to understanding childhood cancer survival in LMICs like Zimbabwe and provide a baseline for future research.

Keywords: childhood cancer; childhood cancer registry; survival; Zimbabwe; sex; performance status.

Introduction

Childhood cancer is not necessarily a health care priority in low- and middle-income countries (LMICs) because of the burden of infectious diseases.^1,2 In addition, there is an inability to accurately quantify cancer incidence because of a lack of dedicated tumour registries.^3,4,5 Childhood cancer represents approximately 1% of all cancers diagnosed annually globally.⁵ Estimates suggest that approximately 85% of all childhood cancer cases occur in LMICs.⁵ Approximately 4.6% of cancers in sub-Saharan Africa occur in children aged 14 years and younger, compared to 0.5% in high-income countries (HICs).⁵

Childhood cancer remains an important cause of paediatric mortality, with an estimated 100 000 cancer-related deaths occurring annually worldwide.⁶ Children diagnosed with cancer in LMICs have an 80% chance of dying compared with approximately 20% in HICs.⁷ Current global estimates indicate large variations in 5-year overall survival (OS) by region, ranging between 8.1% in East Africa and 83% in North America.⁸

A study of three cancer registries (Kampala, Uganda, Abidjan, Cote d’Ivoire, Harare, Zimbabwe) reported a 3-year relative survival rate of less than 60% in children diagnosed with cancer.⁴ Half of the combined paediatric cohort died within 3 years of diagnosis.⁴ Childhood cancer deaths are more common in resource-limited settings because of a lack of access to diagnostic investigations and treatment modalities such as chemotherapy, radiotherapy and surgery, as well as multidisciplinary, high-quality supportive care.⁹ Most childhood cancers can be successfully treated if they are diagnosed early and treated appropriately.¹⁰

KidzCan, a non-governmental organisation (NGO), is assisting Zimbabwe’s Ministry of Health and Childcare in bridging diagnostic, treatment and support gaps in local childhood cancer service delivery.^11,12 KidzCan conducts a wide range of childhood cancer activities, which include awareness campaigns within different communities and provide training to health professionals and traditional medical practitioners.^11,12 KidzCan has a database of children diagnosed with childhood cancer previously or who are currently receiving treatment and support through their organisation.¹² Zimbabwe currently has three paediatric oncologists based at Parirenyatwa Group of Hospitals, a public health facility in Harare, which is also the only paediatric oncology treatment centre in the country.¹² Patients must be referred from primary and secondary care facilities for specialised care at this facility.

The primary objective of this study was to investigate childhood cancer survival 1-year and 2-year post-cancer diagnosis, as well as any associations between survival and age, sex and performance status at diagnosis for children enrolled in the KidzCan database between 2015 and 2021.¹² The secondary objective was to identify determinants of survival for this cohort.

Methods

This retrospective descriptive study analysed KidzCan’s database of children diagnosed with cancer between January 2015 and December 2021.¹² Data collection included age at diagnosis, cancer diagnosis, sex and home address. The cancer diagnosis and stage or risk group (if documented) were recorded from the database and the available hospital records. Malignancies were classified according to the International Classification of Childhood Cancer Third Edition (ICCC-3).¹³ Nutritional status was not documented in the registry and therefore not analysed. However, the captured performance status was included in the data analysis as an indication of the child’s condition at diagnosis. KidzCan self-classified performance status as either ‘active’ (a child able to perform normal daily tasks), ‘frail’ (a child who got tired easily and required assistance with feeding and bathroom routine) or ‘bedridden’ (a child confined to a bed for more than 90% of the day). Outcome data (status and date last seen or date of death) were collected from the hospital records. Follow-up was conducted telephonically for mortality and time to death (Figure 1). Records with a registration date and either a date last seen or date of death or both were included in survival analysis. Records with registration date and date last seen were coded as ‘lost to follow-up’ if the date last seen was before the 2 years had lapsed and were included as censored observations in the survival analysis. Records with a date last seen more than 2 years after registration were deemed alive. Records that had no registration date and no date of death or records that had only one of the two were excluded from survival analysis. The endpoint was defined as OS at 2 years post-cancer diagnosis. For comparison purposes, the OS at 1 year was also calculated.

FIGURE 1: Total cohort (included and excluded groups) with 2-year overall survival.

Data were analysed using SPSS version 28 (IBM SPSS Statistics, United States). Kaplan-Meier curves were used to calculate OS, and log-rank tests were used to compare survival in relation to predictor variables of cancer diagnosis, age, sex, performance status, solid versus haematological malignancies, home region and distance from a treatment facility for the group included in the survival analysis. The chosen predictors were based on known variables that influenced survival after diagnosis.⁵ Cox proportional hazard regression models were used to estimate adjusted hazard ratios (HR) and 95% confidence intervals (CI). Predictors with a univariate p-value < 0.1 were entered as independent variables in the model for time to mortality within 2 years. Predictors with a p-value < 0.05 were retained in the final model. To assess possible bias in the results, the included and excluded cancer case groups were compared in terms of sex, age, cancer type and performance status using Chi-square and t-tests.

Ethical considerations

KidzCan provided consent to access the database used in this study. The Stellenbosch University Health Research Ethics Committee (S22/08/142 PhD) and the Medical Research Council of Zimbabwe (MRCZ/A/2984) granted ethical approval for this study. A waiver of individual patient consent was obtained as the study posed minimal risk to the participants, and data analysis was performed using the de-identified data.

Results

There were 1090 patient records, of which 514 (47.2%) were included for data analysis, and 576 (52.8%) were excluded because of missing data (Figure 1). The 1-year OS for the study cohort was 38% (mean 7.5 months; 95% CI [7.1–8.0]), while the 2-year OS was 17.3% (mean 10 months; 95% Cl [9.7–11.4]) (Online Appendix 1 Figure 1-A1, Online Appendix 1 Figure 2-A1). Male participants had a better 1-year OS of 43.3% (mean 8 months; 95% CI [7.3–8.6]) and a 2-year of 21.9% (mean 11 months; 95% CI [10.4–12.8]) compared to female participants (1-year OS 30.6%, mean 7 months, 95% CI [6.3–7.7], p = 0.016; 2-year OS 10%, mean 9 months, 95% CI [7.8–10.2], p = 0.002) (Figure 2 and Figure 3).

FIGURE 2: Kaplan-Meier 1-year overall survival according to sex at 1 year.

FIGURE 3: Kaplan-Meier 2-year overall survival according to sex at 2 years.

The 5–9-year age group had the best 1-year OS of 40.1% (Online Appendix 1 Figure 3-A1). The 2-year OS decreased to 21.9% in the 5–9-year age group, while the over-15 year age group had no survivors 2 years post-diagnosis (Online Appendix 1 Figure 4-A1). However, age was not associated with survival (1-year OS p = 0.395; 2-year OS p = 0.305).

Solid tumours had the best 1-year OS (42.4% vs 30% for haematological malignancies, p = 0.001) and 2-year OS (17.3% vs 16.3% for haematological malignancies, p = 0.016) (Online Appendix 1: Table 1-A1, Table 2-A1, Figure 5-A1, Figure 6-A1).

The 1-year OS rates were higher than 40% for lymphoma (47.2%), retinoblastoma (46.6%), soft tissue sarcoma (46.4%) and nephroblastoma (44.1%) (Table 1). Leukaemia was associated with a low 1-year OS of 25% (Table 1). The OS decreased to less than 30% 2 years post-cancer diagnosis for all cancer types (Table 2, Online Appendix 1 Figure 7-A1, Online Appendix 1 Figure 8-A1). Cancer diagnosis was associated with OS at 2 years (p = 0.002).

Performance status was associated with OS at 1 year (p < 0.001) and 2 years (p < 0.001). Children with normal activity at diagnosis had a 1-year OS of 55.1%, which declined to 28% at 2 years (Figure 4 and Figure 5). Frail children had a 1-year OS of 31%, which decreased to 11.5% at 2 years. Similarly, bedridden children had the worst 1-year OS of 16.3%, which declined to 13% at 2 years. Patients with unknown performance status had a 1-year OS of 20%, declining to 0% at 2 years (Figure 3). Home region, as an indicator of distance from the treatment facility, was not associated with either 1-year or 2-year OS (p = 0.82 and p = 0.751, respectively) (Online Appendix 1: Figure 9-A1, Figure 10-A1, Figure 11-A1, Figure 12-A1).

FIGURE 4: Kaplan-Meier 1-year overall survival according to performance status at diagnosis.

FIGURE 5: Kaplan-Meier 2-year overall survival according to performance status at diagnosis.

Most deaths (72.8%, n = 118) occurred within the first month of diagnosis. The majority of these children (61.9%; n = 73/118) were frail, of which 20.3% (n = 24/118) were bedridden, 16.1% (n = 19/118) were active at diagnosis and 2.1% (n = 5/118) had an unknown status. Most early deaths (45%; n = 49) occurred in patients with acute leukaemia, followed by nephroblastoma (14.4%; n = 17), lymphomas (5.9%; n = 7) and soft tissue sarcomas (3.4%; n = 4) (Online Appendix 1 Table 3-A1). Within the first year after cancer diagnosis, one-fifth (21%; n = 108) of the cohort was lost to follow-up (Online Appendix 1 Table 3-A1), of which female participants accounted for (54.6%; n = 59/108) compared to male participants (45.4%; n = 49/108). Most children lost to follow-up (62%; n = 67/108) were in the 0–4 years age group (Online Appendix 1 Table 5-A1). Most participants (32.4%; n = 35/108) lived more than 400 km from the treatment facility (Online Appendix 1 Figure 9-A1, Online Appendix 1 Figure 10-A1). The main cancer diagnoses in the lost to follow-up group were retinoblastoma (25.9%, n = 28/108) and nephroblastoma (25%, n = 27/108) (Online Appendix 1 Table 4-A1).

Only sex and performance status were associated with 2-year OS (Table 3). Female participants had a higher risk of dying than male participants (p = 0.003; HR 1.39; CI 1.1–1.7), while those who were frail at cancer diagnosis had a 79% higher chance of dying than those who were active (p < 0.001; HR 1.75; CI 1.4–2.2). Bedridden patients had an 80% chance of dying compared to those who were active (p < 0.001; HR 2.74; CI 1.8–4.1) (Table 3). Neither cancer diagnosis (p = 0.129) nor a group of malignancies (solid vs haematological) (p = 0.092) was associated with OS after adjusting for performance status and sex (Table 3).

The comparison between the included and excluded cases (Online Appendix 1 Table 1-A1) revealed a similar male-to-female ratio (p = 0.474), with the study group being younger (mean age 5 years) (p = 0.002) and containing more haematological malignancies (33.9%; n = 174) than the excluded group (25.9%; n = 149) (p < 0.001) (Online Appendix 1 Table 5-A1).

Discussion

This study documented a 1-year OS of 38%, with a decrease to 17.3% at 2 years. This is worse than a previous study that documented survival rates of less than 40% at 5 years after cancer diagnosis.⁵ The documented variations in survival suggest that the region in which the child lived was a crucial prognostic factor.^8,14 The global 5-year OS rate for childhood cancer is approximately 37.4%.⁸ It is estimated that 90% of the children diagnosed with cancer in Africa will die.⁸ The estimated 5-year survival rates range from 11.6% in Africa, 39.6% in Asia, 55% in Latin America and the Caribbean, 64.4% in Oceania, 74.3% in Europe and 83% in North America.^8,15,16 Previous studies in Harare, Zimbabwe, documented better 3-year survival rates for leukaemia (27.8%), retinoblastoma (34.6%) and nephroblastoma (38.7%) compared to those found in this study.⁵ In this study, the 2-year OS for children with leukaemia was 12.6%, compared to the reported OS of 80% in HICs. European studies documented favourable long-term outcomes, with 5-year OS rates of 96% for children diagnosed with retinoblastoma and 89.4% for those with nephroblastoma, compared to this study’s markedly lower 2-year OS of 28% for retinoblastoma and 20% for nephroblastoma. These findings indicate significant disparities in survival.¹⁷

Sex was a predictor of childhood cancer survival in Zimbabwe, with male patients having a superior survival, which contrasted with the Surveillance, Epidemiology and End Results Program (SEER) data from the United States, where male patients had an inferior survival rate.¹⁸ There is a need to investigate further the influence of socio-economic and cultural factors on male-female survival.¹⁹

Performance status was a risk factor, with the best survival observed in children with normal activity at 2 years (OS 28%). The assumption was that children who were diagnosed early with limited disease had a normal performance status, which might explain their improved survival. More than half (61.9%) of the patients who died within the first month of diagnosis were frail at diagnosis. This probably indicates death because of an advanced underlying disease or treatment toxicity.¹⁰ Late diagnosis of advanced cancer has been reported as the main reason for poor OS in LMICs.⁴ Other factors associated with poor survival included treatment failure, diagnostic delays, treatment abandonment and treatment-related mortality.¹⁰ More community-based awareness-raising campaigns targeting the public and healthcare professionals, coupled with diagnosis-specific referral guidelines, could contribute to reducing diagnostic delays.

Treatment toxicity-related death is a major contributor to mortality in LMICs, with an occurrence rate of 24% – 30% in high-risk patients during the first month of therapy.¹⁰ This study did not investigate toxicity-related deaths because this information was not available. This key factor should be examined in prospective studies. The risk of toxicity-related death depends on the type of cancer, the chemotherapy regimen used and the availability of supportive care. To reduce toxicity-related deaths, an early warning system was validated in Guatemala by St. Jude’s Children’s Research Hospital.²⁰ This system facilitated an early staff response to infections and other causes of rapid clinical decline and reduced the number of children who deteriorated and required intensive care support.²⁰

Improving the quality of care could potentially increase survival by 44.6%, whereas increasing the availability of all treatments could improve survival by 54.1%.⁸ Improvements in these two elements were found to have a superadditive effect on survival.⁸ Initiatives such as the global platform for access to childhood cancer medicines can facilitate the availability and accessibility of quality and affordable childhood cancer supportive care medications for children in LMICs and reduce treatment toxicity.²¹

Treatment abandonment and loss of follow-up are major contributors to treatment failure in LMICs.^5,10,22 Active follow-up of patients in LMIC is common during treatment phases but is not routine after treatment completion.⁵ Caregivers may decide whether to attend post-treatment follow-up clinics based on religious and cultural practices as well as the costs involved, especially when there are no family members living near the treatment facility who can provide accommodation and support.⁵ The high loss to follow-up rates (21% in the first year to 24.9% in the second year) in this study was higher than the abandonment rates reported in Kampala (15%) but lower than those reported in Kenya (54%) and Zambia (46%). This may be attributed to high treatment costs, low educational status, parental religious and cultural beliefs, distance from the treatment facility and cancer diagnosis.^5,10,23 The reasons for abandonment or poor follow-up should be investigated in follow-up studies. To reduce treatment abandonment, patient follow-up systems must be improved. Further research is required to understand the factors that contribute to treatment abandonment in Zimbabwe.

This study highlights the significance of NGOs like KidzCan’s assistance in collecting valuable cancer data, especially survival data, particularly in the absence of active hospital- or national registry-based cancer registries. The results indicate that a broad range of initiatives is required to improve outcomes and bridge the survival gap between HICs and LMICs. These include policy interventions with governments investing in early diagnosis and awareness of multidisciplinary care and service delivery (treatment availability, infection control, nutritional programmes, housing programmes during treatment and follow-up care and reducing the financial burden on families) to improve cancer survival.^5,8,10

In addition, there is a need for further prospective studies investigating the contextual factors influencing loss to follow-up, high early mortality and OS. The evidence generated from these studies can be used to tailor-make contextually appropriate childhood cancer treatment programmes and improve OS.⁹ Greater attention needs to be paid to the quality of care received, which includes access to diagnostic and treatment services, trained personnel and efficient referral pathways that ensure a greater focus on patient needs.⁵ Furthermore, dedicated childhood cancer registries that capture all essential data, including nutritional status, comorbid conditions and abandonment, would be essential to determine the survival and late effects accurately.

This study has several limitations. The database excluded children not registered by KidzCan, especially those who received care in private health facilities, and those who were undiagnosed at the time of the study. Determining patient outcome status was challenging because of missing patient information, and outdated contact information prevented effective patient follow-up. Furthermore, the high loss to follow-up, lack of staging data, nutritional status and tumour characteristics made it impossible to establish the role of these elements in the cohort’s survival. The higher proportion of haematological malignancies in the included group may have influenced the OS because of the lower survival rate compared to solid tumours.

Conclusion

Sex and performance status were key factors affecting overall survival, with 1-year and 2-year rates falling below global benchmarks. To close the survival gap between high- and low-resource settings, coordinated efforts are needed, including early diagnosis, comprehensive care, financial support and long-term follow-up. Further studies on factors like loss to follow-up and early mortality will help develop context-specific strategies to improve outcomes in Zimbabwe.

Acknowledgements

We acknowledge the Board of Directors of KidzCan Zimbabwe for granting access to the KidzCan Database. This article is based on research towards an ongoing PhD by Laura Philip Valerio Sibanda, titled ‘Childhood cancer in Zimbabwe’. The thesis was supervised by Mariana Kruger, Anel van Zyl and Tonya Esterhuizen. The manuscript has been revised and adapted for journal publication. The author confirms that the content has not been previously published or disseminated and complies with ethical standards for original publication.

The authors declare that they have no financial or personal relationships that may have inappropriately influenced them in writing this article.

Laura Sibanda and Mariana Kruger conceptualised the study and developed the protocol. Laura Sibanda collected and analysed the data under the supervision of Mariana Kruger, Anel van Zyl and Tonya Esterhuizen. Tonya Esterhuizen performed the statistical analysis. Laura Sibanda wrote the manuscript. Mariana Kruger, Anel van Zyl and Tonya Esterhuizen assisted with data analysis, critical revision and manuscript editing. Loyce Hlatywayo, Patience Kuona, Vongai Dondo and Daniel Mackenzie contributed to data collection and edited the manuscript. All authors read and approved the manuscript.

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

The views and opinions expressed in this article are those of the authors and are the product of professional research. They do not necessarily reflect the official policy or position of any affiliated institution, funder, agency or that of the publisher. The authors are responsible for this article’s results, findings and content.

References