• acute lymphoblastic leukemia;
  • childhood;
  • trends;
  • cohort study;
  • folate


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Increases in the incidence of childhood acute lymphoblastic leukemia (ALL) have been reported in some countries, while other reports from similar geographical regions have indicated stable rates. The reasons for the discrepancies have been debated in the literature, with the focus on whether the observed increases are “real” or an artifact resulting from improvements in diagnosis, case ascertainment and population coverage over time. We used population-based data from Western Australia to investigate trends in the incidence of childhood ALL between 1960 and 2006. Age-standardized incidence rates (ASRs) and rate ratios (indicating annual percent change) were estimated using Poisson regression. Between 1960 and 2006, the ASR was 3.7 per 100,000 person-years, with an annual percent increase of 0.40% (95% CI: −0.20, 1.00). Between 1982 and 2006, the ASR was 3.8, with an annual percent increase of 0.80% (95% CI = −0.70 to 2.30). This increased to 1.42% (95% CI: −0.30, 3.0) when a sensitivity analysis was undertaken to assess the effect of excluding the final 2 years of data. Annual increases of 3.7% (95% CI: −0.50, 8.00) among children aged 5–14 years, and of 3.10% (95% CI: 0.50, 5.70) in girls, were observed for this latter period. These results were supported by national Australian incidence data available for 1982–2003. There may have been a small increase in the incidence of ALL since 1982 among girls and older children, but an overall increase appears unlikely. No impact of folate supplementation or fortification is apparent. © 2007 Wiley-Liss, Inc.

Acute lymphoblastic leukemia (ALL) is the most common childhood malignancy, accounting for ∼25% of childhood cancer diagnoses each year worldwide.1 Despite extensive research over many years, little is known about the etiology of ALL, although a growing number of theories implicating genetic, in utero and environmental factors are emerging.

Research groups from the UK, Europe, the USA and New Zealand have reported increases in the incidence of childhood ALL over the last 3 to 4 decades, while other research groups from similar geographical regions have reported stable rates.2 There has been considerable debate in the literature as to whether such reported increases are “real,” and possibly due to changes in environmental or demographic risk factors; or whether they represent—at least in part—artifact resulting from improvements in diagnosis, case ascertainment and population coverage over time.2, 3 As cancer reporting is voluntary and national coverage not yet achieved in many of these countries,4, 5 distinguishing between these possibilities is problematic.

In Australia, cancer notification has been mandatory from 1982 and population coverage and case ascertainment of the childhood leukemias are considered to be virtually complete. Furthermore, in Western Australia (WA), we have population-based incidence data on hematological malignancies from 1960.

A study published in 2001 by our colleagues reported a 60% reduction in risk of common ALL among children of WA women who took folate supplements (with or without iron) during pregnancy.6 Periconceptional folate supplementation for the prevention of neural tube defects has been promoted in WA since 1992. In addition, voluntary fortification of some foods (mainly breakfast cereals) has been permitted since 1996. Mandatory fortification of bread making flour with folic acid was recently approved for Australia and New Zealand (June 2007).

The aim of this study was to use population-based Western Australian data to investigate whether the rate of childhood ALL has increased over the past 47 years. We also aimed to investigate whether the promotion of folate supplementation or voluntary fortification of foods may have influenced the incidence of ALL.

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Sources of data on ALL cases

Case notifications were from 2 sources. Information on ALL diagnoses between 1960 and 1981 was obtained from the former Leukemia and Allied Disorders Registry (LADR) of Western Australia. Data had also been collected for 1958–59, but were omitted from our analysis because of concerns about the quality of the data during that period. Cancer notification was not mandatory at that time; however, comprehensive notification was achieved since most of the hematologists, pathologists and clinicians involved in the treatment of lympho-hematopoietic neoplasms were members of the Registry Committee and its Diagnostic Group. The Committee met monthly to collate new cases, for many of which the provisional (unreviewed) diagnosis was reviewed by an expert group. More advanced diagnostic techniques led to discontinuation of the review process after 1975.

There were 200 cases of childhood ALL identified using LADR data for the period 1960–1981; in addition, there were 20 cases recorded as “acute leukemia–not otherwise specified (NOS)” (15 of which had been reviewed by the Diagnostic Group). These NOS cases were enumerated to allow assessment of any potential effect of the level of certainty of diagnosis on changes in incidence over time (see below). Of cases diagnosed from 1960 to 1975, 86.1% had both an “unreviewed” and a “reviewed” diagnosis. Diagnostic review mainly served to classify further the “acute leukemia NOS” group by cell type. Based on the premise that the reviewed diagnosis was more likely to be correct, we used the reviewed diagnosis when it was available and the unreviewed diagnosis for the remaining cases.

Information about 341 cases of ALL diagnosed between 1982 and 2006 was provided by the population-based Western Australian Cancer Registry (WACR), which was established in 1981. There were no cases of childhood “acute leukemia NOS” recorded during this period. At the time of writing this article, the cancer data for 2006 were preliminary.

For comparison with national trends in ALL, incidence data for all of Australia—available from 1982 to 2003—were obtained from the Australian Institute for Health and Welfare (AIHW).7 Cancer incidence data are provided to the National Cancer Statistics Clearing House at the AIHW by all state and territory cancer registries, all of which are members of the Australasian Association of Cancer Registries (AACR). Cancer notifications to the AIHW are mandatory and all efforts are made to check for duplicate cases among contributing registries.

Sources of population data

Population data were obtained from 2 sources. The population denominators for the period 1960–1970 were obtained from the Department of Health, Western Australia. Population data for 1971–2006 were available from the Australian Bureau of Statistics.8 A population census of Australia has been conducted every 5 years from 1971. Intercensal population estimates for the annual resident population were interpolated from the 5-yearly census data.

Acute leukemia NOS cases

We used the proportion of ALL cases among all leukemia diagnoses in WA from the WACR and national Australian data from the AIHW to estimate the proportion of the 20 “acute leukemia NOS” cases recorded on the LADR between 1960 and 1981 that were likely to have been classified as ALL in later years. In the WACR and AIHW data, the proportions of ALL were 80.5% and 79.6% respectively, which were consistent with those from the international literature relating to leukemia data since the 1980s: Europe, 81.2%9; UK, 78.1%10; France, 81.3%11; and New Zealand, 78.7%.12 In the LADR period, 74.6% of all leukemia cases were classified as ALL. When we included the 20 “acute leukemia NOS” cases from the LADR period as ALL cases, the proportion of ALL for this period increased to 81.3%, suggesting that a high proportion of the NOS cases are likely to have been ALL cases. Therefore, we included these 20 cases in the analysis.

Data analysis

Information available for WA cases in this study included date of birth, sex, cancer diagnosis, date of diagnosis and age at diagnosis. Age-specific incidence rates of ALL for age groups 0–4, 5–9 and 10–14 years and age-standardized rates, standardized to the World Standard Population of those aged 0–14 years,13 were calculated for each year from 1960 to 2006. National rates of ALL data for the period 1982–2003 were also calculated using the data from the AIHW. All rates presented are per 100,000 person-years.

Fractional polynomial regression (fracpoly in STATA14) indicated that the best fit for year of diagnosis in relation to ALL incidence was a linear term. Poisson regression (Proc GENMOD in SAS15) was used to estimate the age- and sex-adjusted rate ratios and 95% confidence intervals for year of diagnosis, as well as age- and sex-specific rate ratios for the entire period (1960–2006), and separately for the LADR (1960–1981) and the WACR (1982–2006) periods. The percentage change per year in the incidence of ALL was estimated from these rate ratios.

A sensitivity analysis was performed to examine the effect of excluding data from the first 2 and last 2 years of the study period. This has been recommended when examining changes in incidence rates in relatively small populations.10


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

There were 220 ALL diagnoses in Western Australia during the LADR period (1960–1981) and 341 notified to the WACR (1982–2006). The age at diagnosis was similar in the 2 periods, while the male-to-female ratio appeared to be higher in the WACR period (Table I). Consistent with the known epidemiology of childhood ALL, the rates were higher in males than females and higher in the 0–4 years age group than in the older children. Among males, the rate was somewhat higher in the WACR period than in the LADR period (Table I). Age-standardized rates of ALL in WA between 1960 and 2006 are shown graphically in Figure 1.

thumbnail image

Figure 1. Age-standardized incidence rates of acute lymphoblastic leukemia in Western Australia (1960–2006) and Australia overall (1983–2003). Open circles = Western Australia; Solid squares = Australia.

Download figure to PowerPoint

Table I. Acute Lymphoblastic Leukemia Incidence in Western Australia by Sex, Age Group at Diagnosis, and Specified Time Period
 Total NSexAge-standardized ratesAge-specific rates
Males, N (%)Females, N (%)Male:female ratioTotalMalesFemales0–4 years (N; %)5–9 years (N; %)10–14 years (N; %)
  1. LADR = Leukemia and Allied Disorders Registry; WACR = Western Australian Cancer Registry.

LADR:1960–1981220117 (53.2)103 (46.8) (117; 53.2)2.9 (63; 28.6)1.9 (40; 18.2)
WACR:1982–2006341193 (56.6)148 (43.4)1.303. (182; 53.4)3.1 (99; 29.0)1.8 (60; 17.6)
Whole period:1960–2006561310 (55.3)251 (44.7) (299; 53.3)3.0 (162; 28.9)1.8 (100; 17.8)

The rate ratios—indicative of the percent annual change in incidence—overall, by sex and by age group for specific time periods, are shown in Table II. In both the LADR and WACR periods, the confidence intervals for the rate ratios all included unity. Nonetheless, the results for the WACR period (1982–2006) were suggestive of a 3% annual increase in the older 2 age groups. The annual percent increase for the entire period 1960–2006 was 0.40% (Table II).

Table II. Trends in the Diagnosis of Acute Lymphoblastic Leukemia in Western Australia for Specified Time Periods and by Sex and Age-Group
 LADR: 1960–1981WACR: 1982–20061960–2006WACR: 1982–20041
Cases NRR95% CICases NRR95% CICases NRR95% CICases NRR95% CI
  • LADR = Leukemia and Allied Disorders Registry; WACR = Western Australian Cancer Registry; RR = rate ratio based on Poisson regression; CI = confidence interval.

  • 1

    Data for 2005/06 are excluded for sensitivity analysis.

Year overall2201.0020.981-1.0233411.0080.993-1.0235611.0040.998-1.0103171.0140.997-1.031
Year by sex
Year by age group
 0–4 years1171.0140.985–1.0451820.9890.970–1.0102991.0020.994–1.0111720.9960.973–1.018
 5–9 years631.0030.964–1.043991.0280.999–1.0571621.0080.996–1.020911.0371.004–1.071
 10–14 years400.9650.919–1.014601.0310.994–1.0681001.0020.987–1.017541.0370.995–1.080

When data from the first 2 years (1960 and 1961) or the last 2 years (2005 and 2006) were excluded from the analysis of trend across the entire period, the results remained unchanged (data not shown). However, when data from 2005 and 2006 were excluded from the analysis of the WACR period, the overall annual increase was 1.42%, almost double that observed when those years were included (Table II). These results were also consistent with annual increases of 3.70% among children aged 5 years and older, and of 3.08% in females (Table II). The observed effect of excluding data from the most recent 2 years may be attributable to the lower than average number of cases diagnosed in 2006 (8 compared with a median of 12).

The incidence rate of childhood ALL in Australia between 1982 and 2003 was similar to that observed in WA for the same period (4.1 and 3.9 per 100,000 person-years, respectively). There was a similar pattern in the annual percent increase in the incidence of ALL in WA and Australia in the period 1982–2003, although the increase was greater in WA (1.60% vs. 0.70% in Australia) (Fig. 1; Table III). The annual percent increase in Australia was 1.30% in females and ∼1.46% in children aged 5–14 years old, compared with 3.44% and 3.51–4.05% in WA females and 5–14 year-olds, respectively (Table III).

Table III. Trends in the Diagnosis of Acute Lymphoblastic Leukemia in Western Australia and Australia by Sex and Age-Group, 1982–2003
 Western AustraliaAustralia
Cases NRR95% CICases NRR95% CI
Year overall3021.0160.998–1.0343,2251.0071.002–1.013
Year by sex
Year by age group
 0–4 years1660.9980.974–1.0221,7931.0010.993–1.008
 5–9 years861.0411.005–1.0778911.0161.005–1.026
 10–14 years501.0350.990–1.0825411.0141.001–1.027

The introduction of promotion of periconceptional folic acid supplements in WA in late 1992 and voluntary fortification of food items with folate in 1996 nationally are shown as vertical lines on the age-group-specific trends in ALL between 1982 and 2006 in Figure 2. There was no apparent change in the age-specific rates in ALL in relation to the intake of folic acid.

thumbnail image

Figure 2. Age-specific incidence of acute lymphoblastic leukemia in Western Australia (1982–2006). Open squares = 0–4 years; solid triangles = 5–9 years; open circles = 10–14 years. A = Promotion of periconceptional folate supplementation; B = Voluntary fortification of folate in some foods.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The rate of childhood ALL in WA between 1960 and 2006 was 3.7 per 100,000 person years, similar to other western populations.9, 11, 12, 16, 17 During that period, there may have been a weak upward trend in incidence (0.40% per year), although our results are inconclusive.

As elsewhere, the quality and specificity of cancer diagnosis in Australia have improved over time; accordingly, the WACR data (1982–2006) are likely to be more accurate than the LADR data (1960–1981). The annual increase in the incidence of ALL during the WACR period was 0.80%, double that observed for the entire study period. In addition, the annual percent change in the WACR period almost doubled when the last 2 years of data were excluded, although all these trend estimates were still consistent with no increase. Conducting such a sensitivity analysis has been recommended for studies of rates conducted in relatively small populations,10 because of the potential influence that the rate estimates at the end of the time period can have on the observed trend, and because of the possibility that the end-of-period data are incomplete. The data for Australia were also consistent with a weak increase in the incidence of childhood ALL since 1982.

A number of studies from other countries have reported an increasing trend in the incidence of ALL during the past few decades, while others have reported no change. A significant increasing trend of 0.8% (p < 0.0001) between 1978 and 1997 was reported in Europe by a large collaboration of cancer registries.9 Upward trends have also been reported in Italy between 1977 and 2001 (+1.2%; 95% CI 0.2, 2.3%),16 USA between 1974 and 1991 (+1.6%; 95% CI 0.9, 2.3%),18 New Zealand (+2.2%) between 1968 and 1990,12 and Sweden between 1960 and 1998 (+5.8%; 95% CI 4.7, 6.9).19 Relatively stable rates of ALL have been reported by another Australian study for the period 1982–1991,17 in the UK between 1974 and 1997,10 in Nordic countries between 1982 and 2001,20 and in France between 1990 and 1999.11

A number of reasons have been postulated for the apparent discrepancies in reports from countries that are relatively close geographically, such as Italy and France, or Sweden and other Nordic countries. These include between-country differences in levels of improvement in diagnostic criteria and tools, extent of mandatory reporting and national coverage, completeness of ascertainment over time and time period under study.4, 19, 21 Another possibility is double-reporting of cases registered in more than 1 location as ascertainment improves.

Despite the weak evidence of an overall increase in incidence of ALL in our study, our findings are consistent with a 3–4% increase since 1982 in females and children aged 5–14 years. The national data showed a similar pattern in these groups, but the average yearly increase was lower at 1.3–1.6%. Magnani et al. reported a 3.4% increase in females only in Italy between 1975 and 1998, and a 2.6% increase in the youngest age group.22 Dalmasso et al.,16 Dockerty et al.,12 and Dreifaldt et al.19 all reported the greatest increase in the youngest age group, and Dreifaldt et al. also reported a greater increase among females in that age group.

Studying changes in rates of disease over extended periods of time can assist in elucidating the contributions of putative etiological factors. This is particularly so for childhood diseases where, given the relatively short latency period, changes in environmental causes are likely to influence the incidence of disease in the relative short term. The promotion of periconceptional folic acid supplementation and voluntary fortification of foods with folic acid is one of a few relevant environmental factors known to have changed over time in Australia. However, there did not appear to be an effect of folate on age-specific incidence in this study. It should be borne in mind that there would be a lag in any observed effect of folate intake and that we may not be seeing an impact on ALL rates because only 30% of women have been found to take 200 μg or more daily of folic acid for at least 1 month before pregnancy and the first 3 months of pregnancy.23 At present, it is not known what proportion of women continue to take folic acid beyond the first trimester and for how long. These issues will be investigated in our population-based, Australia-wide case–control study of causes of ALL in children (“AUS-ALL”), which concludes its data collection at the end of 2007.

Our study has some strengths and limitations. The main strength is that we have population-based data for a 47-year time period from high quality sources. The LADR was run by those who managed leukemia treatment in WA between 1960 and 1981; so, cases were unlikely to be missed and most cases underwent a review of their diagnosis by an expert panel. Case notification has been mandatory in Australia since 1982, and the individual state and territory registries have a process in place for checking that cases diagnosed in one state are not also reported in another state or territory. The age-standardized rate of ALL in WA between 1982 and 2003 was similar to that in Australia overall, although the magnitude of the increasing trends in females and children 5 years and older were greater in the WA data. The reasons for these observed differences between WA and Australia may be related to there being fewer cases in WA and therefore greater fluctuation in the annual rates.

The acknowledged uncertainty of diagnoses (as indicated by 20 cases of “acute leukemia NOS”) in the earlier LADR period is a potential limitation. However, we believe that the majority of these are likely to have been ALL cases, since the proportion of all leukemia notifications that were ALL when the NOS cases were added was similar to the proportion of ALL cases in the WACR period, and to that observed nationally in Australia and in other countries.9, 10, 11, 12, 17 Indeed, we believe that our ability to identify these NOS cases, and to determine whether or not they should be included as ALL cases, is a strength of our study. Previous studies that have been unable to ascertain the possible “NOS” status of non-ALL cases in the earlier years are likely to have excluded them, leading to the incorrect reporting of increasing incidence over time.

A potential limitation is the small number of cases diagnosed annually in the WA population, leading to considerable fluctuation of rates over time, and allowing rates from individual years to have a relatively large impact on the observed trends. This is illustrated by the fact that excluding the final 2 years of data (2005 and 2006) led to a doubling of the observed annual increase; this could be related to the preliminary nature of the 2006 data from the WACR.

In conclusion, the age-standardized rate of childhood ALL in WA over the last 47 years is similar to published data from other western countries, but we have not seen the significant increases over time that have been reported elsewhere. However, our findings are consistent with a 3–4% increase in incidence since 1982 in females and children aged 5–14 years. Folate supplementation and fortification have not been accompanied by a decline in incidence of ALL in Western Australia.


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The authors thank Dr. Tim Threlfall at the Western Australian Cancer Registry for extracting the cancer data and Dr. Jim Codde at the Department of Health, Western Australia for providing the population denominatorsx.


  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  • 1
    ParkinD,WhelanS,FerlayJ,TeppoL,ThomasD, eds. Cancer incidence in five continents, vol. VIII (IARC Scientific Publication No. 155). Lyon, France: International Agency for Research on Cancer, 2003.
  • 2
    Maule MM,Zuccolo L,Magnani C,Pastore G,Dalmasso P,Pearce N,Merletti F,Gregori D. Bayesian methods for early detection of changes in childhood cancer incidence: trends for acute lymphoblastic leukaemia are consistent with an infectious aetiology. Eur J Cancer 2006; 42: 7883.
  • 3
    Bukowski JA. Critical assessment of opposing views on trends in childhood cancer. Int J Health Serv 2000; 30: 3737; discussion 79–86.
  • 4
    Adamson P,Law G,Roman E. Assessment of trends in childhood cancer incidence. Lancet 2005; 365: 753.
  • 5
    Steliarova-Foucher E,Stiller C,Kaatsch P,Berrino F,Coebergh JW,Lacour B,Parkin M. Geographical patterns and time trends of cancer incidence and survival among children and adolescents in Europe since the 1970s (the ACCISproject): an epidemiological study. Lancet 2004; 364: 2097105.
  • 6
    Thompson JR,Gerald PF,Willoughby ML,Armstrong BK. Maternal folate supplementation in pregnancy and protection against acute lymphoblastic leukaemia in childhood: a case–control study. Lancet 2001; 358: 193540.
  • 7
    Australian Institute for Health and Welfare. ACIM (Australian Cancer Incidence and Mortality) Books: Canberra: Australian Institute for Health and Welfare, 2007.
  • 8
    Australian Bureau of Statistics. Population by age and sex, Australian states and territories, Preliminary estimates as at 30 June 2006, Cat. No. 3201.0, Australian Bureau of Statistics, 2006.
  • 9
    Coebergh JWW,Reedijk AMJ,de Vries E,Martos C,Jakab Z,Steliarova-Foucher E,Kamps WA. Leukaemia incidence and survival in children and adolescents in Europe during 1978–1997. Report from the Automated Childhood Cancer Information System project. Eur J Cancer 2006; 42: 201936.
  • 10
    Feltbower RG,Moorman AV,Dovey G,Kinsey SE,McKinney PA. Incidence of childhood acute lymphoblastic leukaemia in Yorkshire, UK. Lancet 2001; 358: 3857.
  • 11
    Clavel J,Goubin A,Auclerc MF,Auvrignon A,Waterkeyn C,Patte C,Baruchel A,Leverger G,Nelken B,Philippe N,Sommelet D,Vilmer E, et al. Incidence of childhood leukaemia and non-Hodgkin's lymphoma in France: National Registry of Childhood Leukaemia and Lymphoma, 1990–1999. Eur J Cancer Prev 2004; 13: 97103.
  • 12
    Dockerty JD,Cox B,Cockburn MG. Childhood leukaemias in New Zealand: time trends and ethnic differences. Br J Cancer 1996; 73: 11417.
  • 13
    Waterhouse J,Muir C,Correa P,Powell J. Cancer incidence in five continents, vol. III (IARC Scientific Publications No. 15). Lyon, France: International Agency for Research on Cancer, 1976.
  • 14
    StataCorp. STATA v9.0, ed. 9 College Station, Texas: StataCorp, 2005.
  • 15
    SAS Institute Inc. SAS v.9.1, ed. Cary, North Carolina: SAS Institute, 2003.
  • 16
    Dalmasso P,Pastore G,Zuccolo L,Maule MM,Pearce N,Merletti F,Magnani C. Temporal trends in the incidence of childhood leukemia, lymphomas and solid tumors in north-west Italy, 1967–2001. A report of the Childhood Cancer Registry of Piedmont. Haematologica 2005; 90: 1197204.
  • 17
    McWhirter WR,Dobson C,Ring I. Childhood cancer incidence in Australia, 1982–1991. Int J Cancer 1996; 65: 348.
  • 18
    Gurney JG,Davis S,Severson RK,Fang JY,Ross JA,Robison LL. Trends in cancer incidence among children in the U.S. Cancer. 1996; 78: 53241.
  • 19
    Dreifaldt AC,Carlberg M,Hardell L. Increasing incidence rates of childhood malignant diseases in Sweden during the period 1960–1998. Eur J Cancer 2004; 40: 135160.
  • 20
    Hjalgrim LL,Rostgaard K,Schmiegelow K,Soderhall S,Kolmannskog S,Vettenranta K,Kristinsson J,Clausen N,Melbye M,Hjalgrim H,Gustafsson G. Age- and sex-specific incidence of childhood leukemia by immunophenotype in the Nordic countries. J Natl Cancer Inst 2003; 95: 153944.
  • 21
    Maule MM,Merletti F,Pastore G,Magnani C,Richiardi L. Effects of maternal age and cohort of birth on incidence time trends of childhood acute lymphoblastic leukemia. Cancer Epidemiol Biomarkers Prev 2007; 16: 34751.
  • 22
    Magnani C,Dalmasso P,Pastore G,Terracini B,Martuzzi M,Mosso ML,Merletti F. Increasing incidence of childhood leukemia in Northwest Italy, 1975–98. Int J Cancer 2003; 105: 5527.
  • 23
    Bower C,Miller M,Payne J,Serna P,de Klerk N,Stanley FJ. Folate promotion in Western Australia and the prevention of neural tube defects. Aust N Z J Public Health 2004; 28: 45864.