Discussion

The area served by MTRH suggests that the number of paediatric patients with cancer is many times larger than the annual average of 109 patients who visit the hospital.9,–,11 There are numerous reasons why there are fewer reported than actual cases of childhood malignancies in low-income countries. Cancer cases can only be identified after diagnosis and accurate registration.9 12 Lack of parental education and low awareness of health issues delay seeking of medical help. Parents may not recognise symptoms of paediatric malignancies, or may believe cancer is incurable and therefore do not seek medical care.9 12 13 Parents may rely on traditional healers using herbs or witchcraft.14 Parents may not have a means of transport to a medical centre, or may not be able to pay for diagnostic procedures or treatment.15 16 Healthcare providers may not be skilled in recognising childhood cancer, and laboratory and imaging devices may be lacking.9 12 Personnel may also believe that cancer is incurable or that parents cannot afford treatment and therefore decide referral or treatment is not worthwhile.17 18 And if malignancies are accurately diagnosed, they may not be registered systematically.9 Each misstep can prevent documentation of cases, and results in a lower reported than actual incidence of cancer. These difficulties may be overcome by increasing public awareness of the symptoms, diagnosis and treatment of childhood cancer, emphasising the fact that most cancer can be cured, facilitating access to medical services for poor families, training healthcare providers, enhancing diagnostic capacities and rigorously improving record-keeping.

The distribution of malignancies in Western Kenya is similar to findings in other equatorial African countries but differs markedly from studies performed in high-income countries.12 19,–,25 Such differences may be the result of the cumulative effect of variable genetic predisposition, infectious diseases, environmental exposures and inaccurate registration.26 In high-income countries, haematological malignancies are the most common, followed by brain tumours and solid tumours.9,–,11 Our study reveals that haematological malignancies in Western Kenya are also the most prevalent cancers. However, in high-income countries acute lymphoblastic leukaemia is the most common haematological cancer, whereas in Kenya it is non-Hodgkin's lymphoma. Very few brain tumours are registered at MTRH. The substantial underestimation of brain tumours in low-income countries may be related to scarcity of the neuro-imaging facilities required for diagnosis and limited possibilities to perform tumour biopsies, and requires further exploration in the future.27 The group of solid tumours encompasses the second largest group of cancers in Kenya.

Our study showed that non-Hodgkin's lymphoma was the most common type of cancer in MTRH. In Kenya, Burkitt's lymphoma, a subtype of non-Hodgkin's lymphoma, is much more prevalent than in high-income countries due to increased infection rates with Epstein–Barr virus and immune stimulation by malaria.28,–,31 However, pathologists in MTRH are unable to distinguish Burkitt's lymphoma from other non-Hodgkin's lymphomas because fine needle aspirates are used for diagnosis, facilities for additional staining are lacking and technical challenges hinder the reading of slides.

We found a relatively low rate of acute lymphoblastic leukaemia. The difference between the reported and actual incidence of childhood malignancies in low-income countries is most striking for leukaemia, the symptoms of which resemble those of infections including malaria, and children can die before cancer is suspected or diagnosed. In lymphomas and solid tumours, visible symptoms encourage parents to seek medical help and early death is less common.9

We also found high numbers of retinoblastoma and Kaposi's sarcoma. Non-heritable retinoblastoma has a higher incidence among less affluent populations, suggesting an association with poor living conditions and an infectious aetiology. The high rate of Kaposi's sarcoma in Kenya is due to the high percentage of HIV-positive children.8 32,–,35 In one of our studies at MTRH, we found that 13% of childhood cancer patients were HIV-positive and that 57% of these HIV-positive children were diagnosed with Kaposi's sarcoma (unpublished data).

The absence of Ewing's sarcoma in MTRH is remarkable. In low-income countries a very low incidence of a particular malignancy may indicate systematic misdiagnosis, especially among malignancies that are difficult to identify without expertise and pathological diagnostic facilities. When Ewing's sarcoma, for instance, is often misdiagnosed as rhabdomyosarcoma, the rate of Ewing's sarcomas is strikingly low while that of rhabdomyosarcoma is remarkably high. This necessitates further investigation at MTRH. The diagnosis of certain tumour types may be limited by lack of access to specialised diagnostic stains.9

Our finding that more males than females develop cancer is in accordance with earlier studies in Africa, but is unexplained. It is known that slightly more males have cancer and that boys are more susceptible to Burkitt's lymphoma than girls. Some studies suggest that sex chromosome linked genes may play a role in the latter.12 20,–,23 Other possible explanations for male preponderance are cultural and economic. Parents may consider that boys are more important for the family, contribute more to society and will take care of their aging parents, leading to more frequent presentations of boys than girls to healthcare centres.23 36

Our study shows that there is a wide age distribution of malignancies in Kenya. This pattern is also found in other African countries but differs from that of high-income countries. In high-income countries most childhood cancers occur in children under 5 years of age, decrease between 6 and 14 years of age and increase again between 15 and 18 years of age.10 11 However, the largest group of children attending MTRH are between 6 and 10 years old, perhaps because the majority of children with acute lymphoblastic leukaemia (aged 3–6 years) die of complications before reaching hospital. In addition, parents are aware of the high death rate of children and so if those below 6 years of age fall seriously ill, more parents are resigned to the fact that they will die. However, when children are older they start to contribute to the family income and medical help will be sought sooner. The group of children aged 15–18 years with cancer is quite small.

Gathering data in a low-income country like Kenya posed several difficulties. Filling in of medical records is often not seen as a priority, and storage space is insufficient and frequently changed. Thus medical records were at times mixed up in piles in remote rooms which had not been opened for years. Many medical records were missing, while the records themselves were in poor condition, falling apart and missing notifications because papers had been torn from the files. Important information was either not categorised by date and issue or not documented at all. If medical records are unorganised, the status of patients is very difficult to monitor. Overall, getting access to medical records and extracting reliable data took substantial effort.

Before this study was carried out, information on patients was not gathered in a comprehensive or systematic manner at MTRH. None of the three databases was complete, indicating that the thoroughness and quality of the data sources require profound improvement: for example, all children in whom a malignancy is confirmed need to be recorded in the pathology database.

Our study provides an overview of childhood cancer patients presenting for treatment in Western Kenya over 4 years. We cannot confirm that we succeeded in collecting the names of all childhood cancer patients. However, local caregivers believe that we assembled the vast majority of files and that we are able to provide a reliable overview. Basic data for childhood cancer patients must be collected before improvements can be made. We also learned that more research on non-Hodgkin's lymphoma is advisable. This type of cancer, when diagnosed at an early stage, has a high cure rate, and so treating it successfully would demonstrate that cancer can be cured in Kenya and encourage patients, parents and medical staff. MTRH should continue and extend the new comprehensive cancer registration system to serve as the basis for an evidence-based oncology program, which may lead to better clinical results.9