Pediatric Oncology: Surgical and Medical Aspects Devendra K Gupta, Robert Carachi
INDEX
×
Chapter Notes

Save Clear


1The Scientific Basis of Pediatric Surgical Oncology
zoom view
2

Epidemiology of Childhood TumorsCHAPTER 1

Chantal Rodary
Robert Carachi
Devendra K Gupta
Epidemiology is the study of the frequency and distribution of malignant tumors. The two frequency parameters most employed are incidence and mortality. These will be presented globally for the five continents, and also individually for certain countries, and according to the main characteristics of the subjects (age, sex, etc.) and of the type of cancers, without neglecting the dynamic nature of these parameters over time.
Unlike those found in adults, childhood cancers are particular in many respects: they are rare with the overall cumulative incidence up to 15 years varying only slightly from one country to another (roughly 1.0–2.5 per 1000). The incidence rate standardized on the age distribution of the world population ranges from 75 to 140 per million.1 Furthermore, the histology of these tumors differs considerably and carcinomas are infrequent among the histologic types.
Many childhood cancers have histological features that resemble fetal tissues at various stages of development and are therefore designated as embryonal. Childhood cancers tend to have short latent periods, but are more responsive to chemotherapy than the tumors typically occurring in adults. Prospective epidemiological studies on childhood cancers need to be conducted to allow for uniform treatment policies.
Pediatric tumors should be classified by histology rather than primary site of the tumor. The incidence of childhood cancer is only 2 percent of that in adults in developed countries and about 3 percent in developing countries. It is likely that genetic predisposition has a greater role in the etiology. In the developing countries, reliable data on incidence and mortality of childhood cancers are available from only a few areas. Analysis of specific tumor types show more striking geographic variations of rates that are not readily explained by deficiencies in the data. It is essential that childhood tumors than those of adults, and thus comparisons between different ethnic groups living in the same area or between similar ethnic groups in different environments may be particularly relevant.
In this chapter, frequency parameters, the age of the children (0–14 years), age categories and the classification of tumors are those defined by the International Agency for Research on Cancer (IARC, Lyon, France) in its publications.26
 
CLASSIFICATION OF CHILDHOOD TUMORS
Due to the diversity of the histologic types of childhood tumors, the International Classification of Disease for Oncology (ICDO) used for adults, cannot be applied.7 The Birch and Marsden Classification, recognized by the World Health Organization (WHO) is the classification most often used.8 It is based on histology and the site of the tumors that are divided into 12 diagnostic groups:
  1. Leukemias
  2. Lymphomas and other reticuloendothelial neoplasms
  3. Central nervous system and miscellaneous intracranial and intraspinal neoplasms (including non-malignant tumors, recorded in many cancer registries)
  4. Sympathetic nervous system tumors
  5. Retinoblastomas
  6. Renal tumors
  7. Hepatic tumors
  8. Malignant bone tumors
  9. Soft tissue sarcomas4
  10. Germ-cell trophoblastic and other gonadal neoplasms
  11. Carcinomas and other malignant epithelial neoplasms
  12. Other and unspecified malignant neoplasms
Each of these groups is further subdivided. This classification was revised and modified in 1996 to incorporate recent developments in pathology and epidemiology.9 This work was conducted in collaboration with the IARC and the International Society of Pediatric Oncology (SIOP). The modifications are only applicable to the subgroups. Among them, the subgroup of renal tumors now comprises three categories:
  1. Wilms' tumor, clear-cell sarcoma and rhabdoid tumors
  2. Renal carcinoma
  3. Unspecified malignant renal tumors
The histiocytosis X category, part of group II, has been totally excluded from the classification. Non-malignant intracranial and intraspinal germ-cell tumors are no longer in group III but are now in group X.
A correspondence table has been established with ICD, but this presents some imperfections especially for neoplasms of the nervous system: for instance neuroblastoma is occasionally coded with the organ affected or with connective and soft tissue sarcomas and with the nervous system.10
 
MORBIDITY REGISTRIES
The objective is to compile a registry of all the new cancer cases. If a registry is to be of good quality, it must have an underlying epidemiological structure that conti-nuously and exhaustively records new cancer cases occurring in the population in a given region. These registries are called population-based registries. As childhood cancers are rare, it is important that the population on which the registry is based is sufficiently large to accommodate an adequate number of events permitting reliable estimations.
Currently, few countries are entirely covered by cancer morbidity registries. Denmark was among the first to start a registry (1942), then the Nordic countries (Sweden, Finland, Norway, from 1952), the UK (since 1962) and more recently Canada and Australia (1977). In the USA, the National Cancer Institute (NCI) SEER (Surveillance, Epidemiology and End-Results) program registries cover roughly 10 percent of the population. In African and certain Asian countries, only hospital-based or histopathology-based registries are available and these sources contain information of dubious precision on the size of the population at risk and the number of new cases in the population.
Quality of diagnosis and classification of tumor site also need to be evaluated in each registry, to determine reliability of data.
An International Association of Cancer Registries (IACR) was created in 1966 to support members interested in the development and applications of cancer registration and morbidity survey techniques to studies on well-defined populations. The statistics published by the IARC in 1982 are based on registries of approxi-mately 50 countries. More recent data have been published in other registries.1115
 
Cancer Mortality Data
Since the 1950s, data on mortality have been available in numerous countries. They are published annually for each country by the WHO in the World Health Statistics Directory.
In reality, registration of deaths is not always all-embracing. WHO recommendations concerning registration of deaths are not complied with everywhere and certain procedures can vary between countries often resulting in inaccurate or even unavailable information on the causes of deaths.
 
Survival Data
Some morbidity registries also record data on survival, thus allowing survival rates to be compared by type of cancer. A European study (EUROCARE), which began in 1990, has compiled data from 30 European registries (approximately 800000 patients - of whom just under 8000 are children) to compare survival rates between countries.6
 
MORBIDITY AND MORTALITY PARAMETERS
 
Crude Rate
 
Crude Incidence Rate (IR)
This quantitates new cases over a given period, in a given population. Most often, annual figures are provided per million persons. Data abstracted from the cancer morbidity registries that exist in each country are used to calculate this rate. The denominator, also known as the total number of person-years, is the time accumulated for a subject exposed to a risk of developing a cancer during the period under consideration. The quality of these estimations is obviously contingent upon that of the registries.5
 
Crude Mortality Rate (MR)
This is calculated in a similar manner to that used to calculate the incidence rate, but taking into account the number of deaths. Here, the denominator is the number of individuals at risk of death due to cancer, accumulated for the periods considered (person-years). The calculation of mortality rates is exhaustive, since all the deaths are recorded and the size of the population at risk is counted at the census.
 
Specific Rates
It is possible to calculate specific rates by age categories, site, geographic location, etc. Thus, if we consider age-specific morbidity, four rates, IR0, IR1, IR2, IR3, will be obtained for children aged < 1 year, 1–4 years, 5–9 years, and 10–14 years respectively.
 
Standardized Rates
In order to be able to compare cancer incidence or mortality rates between populations, the effect of confounding factors such as age, sex, site, etc. must be eliminated with standardization methods.
 
DIRECT STANDARDIZATION (STANDARD POPULATION METHOD)
The procedure is based on the calculation of the expected number of cases in each age group of a standard population by applying to the corresponding person-years, the estimated rate of the population under study. The total number of expected cases is then divided by the total number of person-years in the theoretical population and yields an age standardized rate (ASR). This method, used by Parkin can also be used for incidence and mortality rates.2 A cumulated incidence rate (CUM) can also be calculated using the corresponding specific rates for each year by summing the total of each year (0–14).
 
INDIRECT STANDARDIZATION
Standardized incidence rates (SIR) are defined as the ratio between the total number of new cases and the number of new cases expected if the population was subject to the specific incidence rates of the standard population for each age group. The same procedure is applied for the calculation of the Standardized mortality rate (SMR).
 
INCIDENCE RATE FOR TOTAL CHILDHOOD CANCERS WORLDWIDE
The IARC study which essentially covers the decade around 1975, provides the ASR and CUM for certain countries.2 Worldwide ASR vary from 75 to 140 per million population, whereas CUM vary from 1000 to 25000 per million population.
Regarding sex, ASR vary by a ratio of 1:2 (86.3–164.1 per million for boys and 34.5–130.4 for girls). The rates are on average higher among boys than among girls, whatever the country considered.
Concerning race, ASR are higher among whites than among blacks of whatever sex.17
 
Variations in Incidence Rates
The distribution of types of childhood cancers is practically identical worldwide, the most frequent cancers being leukemias (30%) followed by brain tumors and tumors of the central nervous system (20%), lymphomas (14%) and neuroblastomas (8%) (Fig. 1.1). In certain developing countries the relative frequency of lymphomas, retinoblastomas and Wilms' tumors is higher.
zoom view
Figure 1.1: Chart showing frequency distribution of theincidence of pediatric malignancies in percentage
6
The total crude incidence rates in India per 1,000,000 populations were 100 for males and 68 for females.16 Age adjusted incidence rate for world population and crude rates for both the sexes for total childhood cancers did not show any differences. Leukemias showed the highest incidence rate followed by CNS tumors in both the sexes. Male preponderance was observed at all sites and was most marked in the leukemias and lymphomas.
Stiller and Parkin1 have synthesized IARC data having compared the results for countries where the type of cancer was known.2 They found that lympho-blastic acute leukemias, especially during early childhood, are more frequent in countries with a high standard of living and that they represent approximately 80 percent of childhood leukemias in the industrialized countries.
In Hodgkin's disease, the variations between countries and ethnic groups are complex and imply the existence of various etiological factors linked to age or to the histologic subtype. It is well known that Burkitt's lymphoma, identified to have originated in Tropical Africa, is particularly frequent on both sides of the Equator. Socioeconomic factors have been suggested in the causation in these areas. In Nigeria (Ibadan), between 1960 and 1984, almost half of the cancers were Burkitt's lymphomas and the ASR was approximately 80 per million. In high risk areas, Burkitt's lymphoma occur in the head and neck. In most of these countries, the incidence is decreasing as a result of malaria eradication programs. The new histological and immunological data on lymphomas should shed light in the coming years on specific geographical variations of certain forms of lymphoma.
Brain tumors rank second after leukemias in industrialized countries, with the ASR between 20 and 30 per million. In decreasing order of frequency, astrocytomas and medulloblastomas are to be found (included in the 1996 modifications among the primitive neuroectodermal tumors (PNET). It is noteworthy that in developing countries where neurosurgical facilities are lacking and where the number of autopsies is low, the number of new cases is undoubtedly underestimated.
Neuroblastomas account for the quasi-totality of childhood sympathetic central nervous system tumors. During the 1970s, the ASR varied from 7 to 12 per million for populations in Europe, North America and Oceania, with an extremely elevated rate for the first year of life. This rate has risen considerably in Japan since the introduction of early screening in the 1980s: between 1971 and 1980, it varied from 27.1 to 34.1 per million in the first year of life. Since then, it has approached the 100 per million.18 It is widely admitted that many of the neuroblastomas among those screened would never have been clinically detectable and would never have led to fatal oucome.
The retinoblastomas (non-inherited) (which make up 60 percent of the total) are more frequent in the less affluent countries. It has been hypothesized that poor living conditions could be at the origin of infections or that environmental factors may lead to a mutation in utero or during childhood.
The great majority of renal tumors are Wilms' tumors. The highest ASR are observed in black populations regardless of whether Africa or the USA. Age distribution is the same as that of the white populations. The lowest ASR are observed in East Asia.
Hepatoblastoma is one of the least common embryonal tumors whose incidence worldwide is stable (ASR H” 1 per million). Hepatocellular carcinoma is even more rare than hepatoblastoma. It mainly occurs among children suffering from chronic hepatitis B, in regions in Africa or Asia where the rate of liver cancer is elevated among adults.
Concerning bone tumors, slight variations are noted in the incidence of osteosarcomas from one country to another.19 Such is not the case for Ewing's tumors whose incidence rates vary by a ratio of roughly 1:3 with the lowest rates being observed in Africa and in East and South-East Asia and the highest rates in Australia.
Soft tissue sarcomas comprise numerous histologically heterogeneous tumors, the most common among them being rhabdomyosarcoma. France and the Jewish population in Israel have the highest ASR for all of these tumors. The lowest rates are in Asia, but the difference observed may be related to the difficulty in the accurate histologic diagnoses of these tumors where more sophisticated diagnostic techniques are required. Given their limited number, the incidence rates of germ-cell trophoblastic and other gonadal tumors are rather stable. However, the ASR is elevated in East Asia.
 
Worldwide Temporal Trends in Cancer Incidence
The publications are based on the works of the IARC.5,10,20,21 Numerous problems are encountered when attempts are made to establish these estimations over time: rareness of the disease, the calendar period during which these data are available and the grouping together of regions according to geographical or economic development criteria are some of the many factors and sources of heterogeneity that need to be 7taken into account. For instance, Europe is divided into four regions consisting of the following countries: Central Europe (Hungary, Germany (former GDR), Poland, Slovenia); Nordic (Denmark, Finland, Norway, Sweden); Western (France, Germany (former FRG), Italy, Spain, Switzerland); United Kingdom (UK) and Southern Ireland.
In the Nordic region and the UK, the registries cover all the countries concerned, which is not the case for the other regions. To these difficulties must be added the quality of the registries that vary markedly from one country to another.
Coleman uses the linear trend (LT) which is the percentage change in risk every 5 years over the entire data period to express these variations.5 For example, for leukemias, the most frequent cancer in childhood whose diagnosis is the least problematic, he observed the following: the incidence has risen significantly from 1965 to 1985 among boys in central Europe (LT: 3.7%) and for both sexes in the UK (LT: 5.2 and 7.3% respectively for boys and girls). Among girls, it has increased in the Nordic countries (LT: 5.2%) and declined in Western Europe (LT: 22.8%). In Asia and Oceania, although the incidence rates are very heterogeneous, a significant increase is observed among boys in Bombay, Hong Kong and New Zealand (LT respectively: 12.2, 18.1, 8.0%). In the USA, the CUM are stable among blacks as well as among whites.
If these estimations are limited to countries with a population-based registry, the information is more complete and the diagnosis of the type of tumor is more accurate.2225 It should nonetheless be noted that the significant decline in unspecified malignant neoplasms attributable to an increased number of biopsies and progress in histological diagnoses, is at the origin of part of the increase observed.
Gurney found an annual increase of 1 percent in the SIR in the USA (95% confidence interval: 0.6–1.3) between 1974 and 1991 for all the cancers registered by the SEER program.23 This increase is 2 percent per annum or more for tumors of the central nervous system (CNS), rhabdomyosarcomas, germ-cell tumors and osteosarcomas.
Bunin also observed an annual increase of 1 percent for all the cancers registered in the Greater Delaware Valley Registry (USA) between 1970 and 1989.22 This increase essentially concerns CNS tumors (2.7%).
 
MORTALITY RATE
 
Mortality Rate for Total Childhood Cancers Worldwide
Between 1985 and 199010,21 the age standardized mortality rates (ASRM) of the world population were based on WHO mortality data. In Europe they range from 42.4 in Austria to 82.3 in Bulgaria per 1000000 boys and 30.2 in Finland to 59.7 in Portugal per 1000000 girls.
 
Variations in Mortality Rates
 
Role of Cancer in Infant Mortality
If the first year of life is excluded, because of the extent of neonatal mortality after accidents, deaths due to cancer are the second highest cause of death between 1 and 14 years of age.26
 
Temporal Trends in Cancer Mortality
In the worldwide study conducted by Levi based on WHO mortality data a general decline in age standardized mortality rates is observed between 1950 and 1989 for all cancer.10,21 Temporal trends vary from one country to another. Thus in Europe, these declines were generally earlier and larger in Northern Europe compared with Southern Europe, and mostly with Eastern European countries; this reduction in mortality is still evident over the most recent calendar periods. The range of variation in total childhood cancer mortality was around a factor of 2 in both sexes, with the highest rates in Bulgaria, Portugal, Hungary, Czechoslovakia and Poland, and the lowest rates in Austria, UK, Germany, The Netherlands and Finland.
In the majority of European countries, these declines are sharp for leukemias, Wilms' tumors, Hodgkin's disease and other lymphomas. It is more difficult to interpret data from the American continent or Oceania because of the dubious reliability of death certificates in certain countries in Latin America and Asia. It can however be stated, beyond reasonable doubt, that the highest mortality rates for all types of cancer are in certain Latin American countries, in Kuwait, New Zealand and Singapore (between 65 and 75 per 1000000 among boys and between 50 and 65 per 1000000 among girls). When compared with the developed countries where mortality rates are low: Canada, USA, Australia, Japan and Israel (roughly 45 per 1000000 8among boys and between 35 and 40 per 1000000 among girls) the ratio between sexes is 1:8.
In a study from developing countries, the death rate per 1,000,000 population in pediatric malignancies in decreasing order of death rate was leukemia (20), CNS tumors (10), lymphoma (5) and others (1 each).16
A study concerning the entire US population provides accurate results for the period 1950–80. For the period 1965–79, standardized mortality rates were obtained by using data for the 1950–54 period as a reference. The decline in mortality was marked from 1965, and during 1965 through 1979, the number of deaths observed was 44 percent below the expected number at the 1950 rate. With respect to the type of tumor, the decline in mortality was 50 percent for leukemia, 32 percent for non-Hodgkin's lymphoma (NHL), 80 percent for Hodgkin's disease, 50 percent for bone sarcomas, 68 percent for kidney cancer and 31 percent for all other cancers.
 
SURVIVAL RATES
Recorded survival data allow a more detailed assessment of the temporal evolution of cancers.17,2732
Among these studies, Stiller's report analyzed 5-year survival rates in the UK in a series of 15000 children suffering from a cancer diagnosed between 1971 and 1985, with a follow-up of at least 3 years.30 A notable improvement was observed in survival rates except for those regarding brain tumors (excepting medullo-blastomas), Ewing's tumors, fibrosarcomas, neuro-fibrosarcomas and other soft tissue sarcomas for which progress is minimal. The most spectacular improvements (as regards the c2 test for linear trends) between 1971–73 and 1983–85 concerned ALL (acute lymphoblastic leukemia), NHL and neuroblastomas (localized forms). They were also marked for ANLL (acute non-lymphoblastic leukemia), rhabdomyosarcomas and Wilms' tumors.
In 1994, Stiller continued the same study including children suffering from a cancer diagnosed between 1980 and 1991, in order to assess the impact of recent developments in medical care since 1985.31 Compared with the results published in 1990, the notable findings since 1985 were an improvement in ANLL and in short-term (1-year) survival of Ewing's sarcomas. No progress was observed for neuroblastomas. Stiller concluded: ‘The projections of 10-year survival for children who had cancer diagnosed in 1989–91 suggested an overall increase in survival of 19 percent since 1980–82.’ This suggests that nearly two-thirds of children who have cancer diagnosed can expect to survive at least 10 years.31
The significant advances achieved during the past 30 years are due to the multidisciplinary care, the employment of new treatments administered within the framework of standardized protocols, and in hospitals specializing in childhood cancer. The creation of an international organization such as the International Society of Pediatric Oncology in 1969 and the establishment of different cooperative groups have permitted better coordinated research programs (Children's Cancer Group, Pediatric Oncology Group in the USA and United Kingdom Children's Cancer Study Group in the UK).
Further progress will undoubtedly ensue through analytical and experimental epidemiology. To this end, more reliable descriptive data are required. The problems with the reliability of the diagnosis on death certificates and completeness of data have been alluded to in this chapter. Draper underscored the risks of misinterpreting the results of mortality data.33
  1. Tumor can be misclassified at diagnosis (e.g. neuroblastoma when using the ICDO code).
  2. Random fluctuations (especially if rates are based on small numbers).
  3. Trends in mortality may be the results of, or may be obscured by underlying incidence rates.
If epidemiology is to fulfill its objectives then accurate data of good quality should be obtained and studies should use population-based registries as a source of solid, reliable data.
 
AGE AND SEX DISTRIBUTION
Most of the tumors in children occur more frequently in boys than girls. However, for acute nonlymphatic leukemia, osteosarcoma, retinoblastoma and melanoma the sex ratio is around unity. The tumors more frequent in girls are Wilms' tumor, carcinomas of the adrenal cortex and thyroid.
Tumors more common in infants include neuroblastoma, retinoblastoma, soft tissue sarcoma and hepatic tumors. The incidence of hepatic tumors is highest in children aged under 1 years. Acute lymphatic leukemia is well documented for distinctive peak in the age range of 1–5 years. Wilms' tumor and germ cell tumor are more common in children under 5 years than in older age groups.9
Wilms' tumor comprises approximately 95 percent of all renal neoplasms. Sixty five percent Wilms' tumors occur in children under 5 years of age, 35 percent in children aged 5–9 years and 10 percent in 10–14 children aged years.
Osteosarcoma is very rare below the age of 5 years, but increases steeply thereafter. Ewing's sarcoma is also very rare under 5 years and the incidence increases with age, but less markedly than for osteosarcoma.
Retinoblastoma has the lowest median age of all childhood cancers (approximately 15 months) and bilateral cases tend to be diagnosed at a younger age than unilateral cases. Incidence peaks in the first year of life and declines gradually with age thereafter.
To summarize, childhood cancers are much more in boys than in girls. Leukemias are the most common cancers affecting children followed by CNS tumors and lymphomas. Fatality from cancer in children appears to be of the same order in both the sexes. There is no significant change in incidence patterns of childhood cancer worldwide and over time.
REFERENCES
  1. Stiller CA, Parkin DM. Geographic and ethnic variation in the incidence of childhood cancer. Brit Med Bull 1996;52:682–703.
  1. Parkin DM, Stiller CA, Draper GJ, et al. (Eds). International Incidence of Childhood Cancer. IARC Scientific Publications  Lyon,  France: No 87, 1988.
  1. Parkin DM, Stiller CA, Draper GJ, et al. The international incidence of childhood cancer. Int J Cancer 1988; 42: 511–20.
  1. Parkin DM, Muir CS, Whelan SL, et al. Cancer Incidence in Five Continents Vol 6. IARC Scientific Publications  Lyon,  France: No 120, 1992.
  1. Coleman MP, Estève J, Damiecki P, Arslan A, Renard H. Trends in Cancer Incidence and Mortality. IARC Scientific Publications  Lyon,  France: No 121, 1993.
  1. Berrino F, Sant H, Verdecchia A, Capocaccia R, Hakulinen T, Estève J. Survival of Cancer Patients in Europe. The EUROCARE Study. IARC Scientific Publications  Lyon,  France: No 132, 1995.
  1. World Health Organization. International Classification of Diseases for Oncology. WHO,  Geneva:  1976.
  1. Percy C, Van Holten V, Muir C. ICD-0. International Classification of Diseases for Oncology (2nd edn). World Health Organization,  Geneva:  1990.
  1. Kramarova E. The international classification of childhood cancer. Int J Cancer 1996; 68: 759–65.
  1. Levi F, La Vecchia C, Lucchini F, Negni E, Boyle P. Patterns of childhood cancer incidence and mortality in Europe. Eur J Cancer 1992;28A:2028–49.
  1. Gurney JG, Severson RK, David S, Robison LL. Incidence of cancer in children in the United States: sex-, race-, and 1-year age-specific rates by histologic type. Cancer 1995;75:2186–95.
  1. Stiller CA, Allen MB, Eatock EM. Childhood cancer in Britain: The National Registry of childhood tumours and incidence rates 1978–1987. Eur J Cancer 1995;31A:2028–34.
  1. Kaatsch P, Haaf G, Michaelis J. Childhood malignancies in Germany. Methods and results of a nationwide registry. Eur J Cancer 1995;31A:993–9.
  1. Ross JA, Severson RK, Pollock BH, Robison LL. Childhood cancer in the United States. A geographical analysis of cases from the Pediatric Cooperative Clinical Trials Groups. Cancer 1996; 77:201–7.
  1. McWhirter WR, Dobson C, Ring I. Childhood cancer incidence in Australia. 1982–91. Int J Cancer 1996;65:34–8.
  1. Yeole BB, Advani SH, Sunny Lizzy. Epidemiological Features of Childhood Cancers in Greater Mumbai. Indian Pediatrics 2001;38:1270–7.
  1. Miller BA, Ries Lag, Hankey BF, et al (Eds): SEER Cancer Statistics Review: 1973–1990. National Cancer Institute. NIH Pub 1993; 92: 2789.
  1. Sawada T, Kidowaki T, Sugimoto T, Kusunaki T. Incidence of Neuroblastoma in infancy in Japan. Med Pediatr Oncol 1984;12:101–3.
  1. Parkin DM, Stiller CA, Nectoux J. International variations in the incidence of childhood bone tumors. Int J Cancer 1993;53:371–6.
  1. Draper GJ, Kroll ME, Stiller CA. Childhood cancer. In: Doll R, Fraumeni JF Jr, Muir CS (Eds): Trends in Cancer Incidence and Mortality. Cancer Surveys 19/20. Cold Spring Harbor Laboratory Press  New-York:  1994;493–517.
  1. Levi F. La Vecchia C, Lucchini F, Negni E. Boyle P. Patterns of childhood cancer mortality: America, Asia and Oceania. Eur J Cancer 1995;31A:771–82.
  1. Bunin GR. Increasing incidence of childhood cancer: report of 20 years experience from the Greater Delaware Valley Pediatric Tumor Registry. Paediatr Perinat Epidemiol 1996;10:319–38.
  1. Gurney JG, Davis S, Severson RK, Fang JY, Ross JA, Robison LL. Trends in cancer incidence among children in the US. Cancer 1996;78:532–41.
  1. Blair V, Birch JM, Patterns and temporal trends in the incidence of malignant disease in children. Leukaemia and lymphoma. Eur J Cancer 1994;30A:1490–97.
  1. Blair V, Birch JM. Patterns and temporal trends in the incidence of malignant disease in children II. Solid tumours of childhood. Eur J Cancer 1994;30A:1498–1511.
  1. Higginson J, Muir CS, Munoz N. Human Cancer: Epidemiology and Environmental Causes. Cambridge University Press,  Cambridge:  1992;58:439–96.
  1. Miller RW, McKay FW. Decline in US Childhood cancer mortality, 1950 through 1980. J Am Med Assoc 1984;251:1567–70.
  1. Young JL, Gloeckler Ries L, Silverberg E, Horm JW, Miller RW. Cancer incidence, survival and mortality for children younger than age 15 years. Cancer 1986;58(Suppl 2):598–602
  1. Birch JM, Marsden HB, Morris Jones PH et al. Improvements in survival from childhood cancer: Result of a population based survey over 30 years. Brit Med J 1988;296:1372–6.
  1. Stiller CA, Bunch KJ. Trends in survival for childhood cancer in Britain diagnosed 1971–85. Brit J Cancer 1990;62:806–15.
  1. Stiller CA. Population based survival rates for childhood cancer in Britain, 1980–91. Brit Med J 1994;309:1612–16.
  1. Adami HO, Glimelius B, Sparen P, Holmberg L, Krusemo UB, Poten J. Trends in childhood and adolescent cancer survival in Sweden: 1960 through 1984. Acta Oncol 1992;31:1–10.
  1. Draper GJ. Childhood cancer: Trends in incidence, survival and mortality. Eur J Cancer 1995; 31A:653–4.
  1. Birch JM, Marsden HB. A classification scheme for childhood cancer. Int J Cancer 1987;40:620–4.