![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
European Centre on Health of Societies in Transition and * Medical Statistics Unit, London School of Hygiene and Tropical Medicine, London, WC1E 7HT, UK
3To whom correspondence should be addressed. E-mail: Joceline.Pomerleau{at}lshtm.ac.uk.
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
KEY WORDS: • fruit • vegetables • diet surveys • food supply • epidemiologic research design
The 2002 World Health Report described the results of the WHO global burden of disease (GBD)4 2000 study (1). The GBD project estimated the burden of premature death and disability attributable to 26 major risk factors, including low fruit and vegetable intake. The findings suggest that increasing individual fruit and vegetable consumption could contribute to reducing the worldwide burden of disease for ischemic heart disease and ischemic stroke by 30 and 19%, respectively (2). For stomach, esophageal, lung, and colorectal cancer, the potential reduction in disease attributable to an increase in fruit and vegetable intake is estimated to be 19, 19, 12, and 2%, respectively.
Clearly these estimates are subject to the limitations of the Comparative Risk Assessment (CRA) method used in the GBD Project (1,3), including constraints related to the availability, reliability, and validity of data on the worldwide distribution of exposure levels, i.e., fruit and vegetable intakes. Because the effect of low fruit and vegetable intake was examined for the first time, it is important to discuss the difficulties involved in estimating the intake distributions required for the CRA, especially because, as with many risk factors, comprehensive global data on intake do not exist and those that do are often difficult to locate.
The objective of this paper was therefore to summarize the methods used to estimate exposure inputs required for the CRA and their limitations, as well as provide estimates of fruit and vegetable intake stratified by subregion, age, and sex. Full details of the methods and results are available elsewhere (2). The process is inevitably imperfect and future research needs are discussed.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
The CRA methodology required worldwide estimates of fruit and vegetable consumption stratified by 14 subregions (191 countries), 8 age groups, and gender. Countries were grouped into 5 mortality strata on the basis of combinations of child (<5 y old) and adult (15–59 y old) mortality (1). These mortality strata were then applied to the 6 main WHO regions (Africa, Americas, Eastern Mediterranean, Europe, South-East Asia, and Western Pacific) to produce the 14 epidemiological subregions (see Table 1).
|
Dietary survey data. These were identified from a systematic search of computerized databases of published articles, library catalogues, hand-searching of bibliographies, internet searches, and contact with over 150 experts. Selection criteria included the following: 1) recent study (1980 onwards); 2) reference population described; 3) valid sampling strategy; 4) representative sample; 5) sample size large (sample size calculations ideally included); 6) wide age range; 7) nonresponse documented; 8) population-based cross-sectional studies or baseline assessment of large cohort or intervention studies on representative samples; 9) limited identifiable sources of bias; 10) individual data; 11) appropriate statistical analyses; and 12) data available as grams of daily fruit and vegetables. Estimates had to exclude potatoes, pulses, and starchy vegetables and fruits to be consistent with current international recommendations (6); they included fruit and vegetable juices.
Survey data were obtained for 26 countries within 9 subregions.5 In most cases, original data were reanalyzed and provided (see acknowledgments) according to CRA categories. Where necessary, systematic extrapolations were made, as follows (2).
Use of FAO availability statistics. For the 5 subregions for which no survey data were obtained, FAO data on availability of fruits and vegetables (2,9) were used to obtain "FAO-derived proxy mean intakes." Although discrepancies exist between the FAO and survey data (10), this is likely to provide better estimates than would the application of data from other subregions with different food supply patterns (11–13). FAO data on fruit (excluding wine: FAO code 2919) and vegetable (excluding potatoes: FAO code 2918) availability and population size estimates were used (5) to calculate population-weighted mean fruit and vegetable per capita availability for the period 1996–1998 for each subregion. A comparison of FAO and survey data in less-developed countries, in which both were available, yielded inconsistent results; therefore, no correction factor was used.
Because food balance sheets do not provide information stratified by gender and age, a 2-step approach was then used. Step 1: using available survey data (9 subregions), the proportion of total fruit and vegetable intake consumed by different age/sex groups in each subregion was calculated; as expected, the distributions of intakes were strongly influenced by the population structures of the subregions. Step 2: for each subregion with no survey data, the distributions of intakes (Step 1) of the subregion displaying the most similar population structure ("proxy" subregion) were applied to the FAO per capita estimates. This generated "FAO-derived proxy mean intakes" by age and sex. The pooled SDs of the subregion displaying the closest mean intakes were then applied.
Statistical methods
Pooling of survey data. Survey data were pooled statistically6 assuming that each subregion is a stratified sample with the strata being countries, the pooled mean intakes and SDs are representative of the true estimates for a subregion, and differences between the pooled estimates and figures for countries in which data are not available would tend to cancel each other out. If there is substantial heterogeneity among countries in a subregion, the methods used tend to underestimate the true standard error of the pooled mean and SD. Using data for only a few countries may also underestimate the true variation of intakes within a subregion. However, for most subregions, data were available for at least 2 countries, and in 7 subregions, a large proportion of the total population was covered by the surveys (America A = 87.5%; America B = 32.0%; Europe A = 71.3%; Europe C = 69.2%; South East Asia D = 93.7%; Western Pacific A = 97.6%; Western Pacific B = 84.0%).
When data were available for only 1 country within a subregion, 2 approaches were used. For America A and the Western Pacific Region B, available surveys were in countries representing 84–88% of the total population of the subregions (the United States and China, respectively). We thus assumed that data from these countries were representative of subregional intakes. However, for the Eastern Mediterranean Region B and Europe B, data were from countries representing only a very small proportion of the total subregional population (1.4% for Kuwait and 3.8% for Bulgaria). Here a different approach was adopted, involving pooling survey and FAO food balance sheet data. "FAO-derived proxy intakes" were calculated for these 2 subregions using the methods described above but here applying a correction factor to the FAO data because food balance data tend to overestimate intakes in developed countries (14). The correction factor corresponded to the discrepancy between the 2 sources in the subregions with the most similar fruit and vegetable supply (America A for the Eastern Mediterranean Region B, and Europe C for Europe B). We assumed that intakes by gender and age were similar to those observed in Kuwait and Bulgaria.
Distribution of intakes. Due to the lack of available information on the distributions of intakes from survey data, we assumed that they were normally distributed. However, dietary intakes are typically skewed toward higher values (15). Assuming a normal distribution creates the problem that some individuals are recorded as having negative consumption when estimating impact fractions in the GBD study. Because this is impossible, the normal distribution was truncated at zero, and those below this value were allocated a value of zero. A sensitivity analysis was performed to evaluate the likely effects of the normality assumption on the impact fractions (2). This analysis was based on data from the United States for which skewness estimates were available. The results suggested that the approach used gives conservative estimates of attributable fractions for the global burden of disease estimates.
As with other elements of the GBD program, all data and methods used and summarized in this paper were reviewed by anonymous external international reviewers selected by the WHO. This paper presents some CRA results; CIs for the mean and SD estimates can be found elsewhere (2).
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
Estimates of SDs (Table 3) varied considerably by region, gender, and age group, with an overall median SD of 223 g/d. SDs tended to be lower in females than in males on average (but with variations by age group), and they were generally lower in young children. In some subregions, SDs were slightly smaller in the elderly.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
One major limitation in the assessment of exposure to the risk factor "low fruit and vegetable intake" is that the collective term "fruit and vegetables" covers a very heterogeneous group of foods that differ among countries and cultures. Even in a "typical" Western diet, this group includes a wide variety of roots, leaves, stems, fruit, and seeds that varies over time and place (19). The food content of potentially important dietary components depends on numerous factors such as preparation method, variant of product, growing conditions, and storage conditions, a factor of increasing importance as commodities are transported globally to ensure year round supply in industrialized countries. In spite of this complexity, we kept fruit and vegetables as a single entity largely because uncertainty remains concerning which components of fruit and vegetables would confer a beneficial effect (20). However, this highlights the need for international agreement on the definitions of the fruit and vegetable food group. Definitions should take into account regional diversity, agricultural production, as well as current knowledge of the health benefits of fruit and vegetables. Second, pragmatically, obtaining intake data for specific foods would have been even more difficult than for fruit and vegetables taken together.
Due to the lack of information on seasonal variations in intakes, we had to assume that estimates represent long-term annualized mean fruit and vegetable intakes and that they predict disease risk. This assumption requires caution because the pattern of consumption of fruit and vegetables is known to remain seasonal in many countries in spite of the spread of the global economy (e.g., annual cycle of seasonal excesses and out-of-season shortages in the less economically developed countries of the former Soviet Union), with evidence that seasonal shortage may contribute to cardiovascular disease (21). Patterns of intake may indeed be of prime importance as is illustrated by the case of alcohol, for which the risk of cardiovascular disease appears to be more sensitive to the pattern of alcohol consumption over time as well as the total amount consumed (22–24).
Another limitation of the study relates to the lack of survey data from many regions of the world and, when data were available, to the intrinsic difficulties of survey methodology and food intake assessment, and to the fact that the methods used differed among countries. Survey sampling is prone to bias (25), and it is possible that survey respondents were not entirely representative of the reference populations, even though most data were from national surveys of dietary intakes. For example, a lower socioeconomic status is associated with both lower intakes of fruits and vegetables in Europe (26) and lower response rates in health surveys (27). Dietary assessment is also affected by difficulties that include, among others, the conversion of food frequencies into mean intakes in surveys that used FFQ, and the limitations and completeness of the various computerized food analysis software used in different countries (28). In addition, the quality and validity of dietary data collected at the individual level depend directly on the ability and willingness of respondents to provide accurate information on their intakes (29,30). The high fruit and vegetable intakes observed in Europe A and Western Pacific A subregions may represent true elevated intakes. However, they may also suggest conscious or unconscious overreporting of intakes by survey respondents (31); caution is required when making this assumption, however, because little is known of social desirability bias in this field or of its potential social and cultural determinants. The reported intakes in some countries within these subregions were indeed greater than expected (particularly in the United Kingdom and Germany). It is possible that recent public health campaigns, such as those that took place in Finland (32) coupled with changes in retail trade, marketing, and distribution of fruit and vegetables, might have helped improve dietary habits in these populations, in line with the striking improvements in cardiovascular mortality that they have experienced. Conversely, it is possible that the inclusion of fruit juices in the estimates of intakes in this study made the estimates appear larger than might be expected. Future attempts to quantify fruit and vegetable intake should thus ideally try to examine juices separately. More generally, our experience with the estimation of worldwide fruit and vegetable intake clearly shows the urgent need for the development of standardized approaches and tools to collect valid and comparable information on fruit and vegetable intake in more developed and developing countries.
In addition to their influence on mean intake estimates, survey methods are likely to have affected our estimates of variation in intake, i.e., SDs. Because most surveys identified for this study collected data with one 24-h recall, it is expected that SDs were artificially increased due to large intraindividual variation in intakes (4). However, the statistical methods used to pool data from 2 or more surveys tend to underestimate the level of uncertainty surrounding the pooled SD for a subregion based on a subsample of countries if there is substantial between-country variation. Future research should investigate further daily variations in fruit and vegetable intake and their influence on estimates of SD in various populations throughout the world.
Because food availability statistics were used for subregions with no or few data available, some estimates may also have been influenced by the sources of uncertainty affecting food balance sheet data. These include, among others, the availability and validity of the basic national information on which they are based, and these are known to vary among countries, and from 1 year to another, both in terms of coverage and accuracy (9). It has been reported that estimates of the net availability of vegetables is complicated in many countries by factors such as noncommercial production and uncertain losses to animal feed, spoilage, and waste. However, the FAO performs external consistency checking using supplementary information. In this study, 3 years of data were used to reduce the effect of potential yearly variations in coverage and accuracy. Our methods to derive intakes from availability statistics also have major limitations, including the current lack of information on appropriate adjustment factors and the influence of the population structure of the subregions on the estimates of intake distribution among genders and age groups. Differences in population structures among subregions could make the estimates less reliable, particularly in population strata with relatively smaller sample sizes such as in the elderly. Finally, a major disadvantage of food balance sheet statistics is that they could not provide information on variations in intakes.
The approaches used to obtain our final subregional estimates, including the extrapolations and assumptions made when pooling data and when combining survey and FAO data, are further potential sources of error in this study. It is also possible that the current grouping of countries did not reflect well the heterogeneity of exposure and disease experience (e.g., combining northern European and Mediterranean countries although they have very different disease and intake patterns).
A way to assess the accuracy of our regional estimates of mean intakes may be to compare them with regional mean levels of biomarkers of fruit and vegetable intake. However, levels of biomarkers have to be interpreted cautiously (33). In any case, the availability of such data is currently too limited to allow for any comparison with intake data.
Given so many limitations, there were 2 options with regard to the assessment of the global burden of disease attributable to low fruit and vegetable intake, i.e., either exclude regions without good exposure level data (which would have meant that the focus of nutritional epidemiological research would continue to be concentrated on developed countries) or use clear assumptions and extrapolations that would stimulate the need for better data collection and further research in all regions. We believe that the estimates provided are the most extensive data set available derived from the best evidence currently available. They should be interpreted with caution and treated as a first attempt to develop methods that can be used to assess the worldwide burden of disease due to low fruit and vegetable intake. It is hoped that by confirming the importance of dietary factors for the burden of disease worldwide, the results from the GBD study update will encourage more countries to assess the dietary intake of their population using comparable methods of data collection, agreed through international consensus that could meet the objectives of each country and balance the needs for simplicity, low costs, and greatest accuracy. This will stimulate further discussion about the growing importance of estimating the share of the global burden of disease attributable to nutritional factors.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Supported by a grant from the English Department of Health. However, the English Department of Health cannot accept responsibility for any information provided or views expressed.
4 Abbreviations used: AFR-D, Africa D subregion; AFR-E, Africa E subregion; AMR-A, America A subregion; AMR-B, America B subregion; AMR-D, America D subregion; CRA, Comparative Risk Assessment; EMR-B, Eastern Mediterranean B subregion; EMR-D, Eastearn Mediterranean D subregion; EUR-A, Europe A subregion; EUR-B, Europe B subregion; EUR-C, Europe C subregion; GBD, global burden of disease; SEAR-B, South East Asian B subregion; SEAR-D, South East Asian B subregion; WPR-A, Western Pacific A subregion; WPR-B, Western Pacific B subregion.
5 The list of 26 countries and details of the dietary intake studies used are available with the online posting of this article at www.nutrition.org.
6 Further details of the statistical methods used for pooling dietary survey data are available with the online posting of this article at www.nutrition.org.
Manuscript received 25 November 2003. Initial review completed 30 December 2003. Revision accepted 12 February 2004.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
1. World Health Organization (2002) The World Health Report 2002. Reducing Risks, Promoting Healthy Life 2002 WHO Geneva, Switzerland.
2. Lock, K., Pomerleau, J., Causer, L. & McKee, M. (2004) Global burden of disease due to low fruit and vegetable consumption. Ezzati, M. Lopez, A. D. Rodgers, A. Murray, C.J.L. eds. Comparative Quantification of Health Risks: Global and Regional Burden of Disease Due to Selected Major Risk Factors 1 World Health Organization Geneva, Switzerland. (in press).
3. Ezzati, M., Lopez, A. D., Rodgers, A., Vander Hoorn, S. & Murray, C.J.L., and the Comparative Risk Assessment Collaborating Group (2002) Selected major risk factors and global and regional burden of disease. Lancet 360:1347-1360.[Medline]
4. Cameron, M. & van Staveren, W. (1988) Manual on Methodology for Food Consumption Studies 1988 Oxford University Press Oxford, UK.
5. FAOSTAT () (Online. Food and Agriculture Organization. http://apps.fao.org/) [accessed Dec. 1, 2001].
6. World Cancer Research Fund and American Institute for Cancer Research (1997) Food, Nutrition and the Prevention of Cancer: A Global Perspective 1997 American Institute for Cancer Research Washington, DC.
7. Food and Agriculture Organization, World Health Organization & United Nations University (1985) Energy and Protein Requirements. Report of a Joint FAO/WHO/UNU Expert Consultation. Technical Report Series no. 724 1985 WHO Geneva, Switzerland.
8. World Health Organization (1990) Diet, Nutrition and the Prevention of Chronic Diseases 1990 WHO Geneva, Switzerland.
9. Kelly, A., Becker, W. & Helsing, E. (1991) Food balance sheets. Becker, W. Helsing, E. eds. Food and Health Data. Their Use in Nutrition Policy-Making 1991:39-47 World Health Organization Regional Publications, European Series, No. 34. WHO Regional Office for Europe Copenhagen, Denmark. .
10. Pomerleau, J., Lock, K. & McKee, M. (2003) Discrepancies between ecological and individual data on fruit and vegetable consumption in fifteen countries. Br. J. Nutr. 89:827-834.[Medline]
11. World Health Organization (2000) The World Health Report 2000 2000 WHO Geneva, Switzerland.
12. World Bank (2000) Classification of economies [Online] 2000 . World Bank http://www.worldbank.org/data/databytopic/class.htm#Low_income [accessed Dec. 1, 2001].
13. Directorate of Intelligence (2000) World Fact Book 2000 2000 Central Intelligence Agency Washington, DC.
14. Joffe, M. & Robertson, A. (2001) The potential contribution of increased vegetable and fruit consumption to health gain in the European Union. Public Health Nutr. 4:893-901.[Medline]
15. Willett, W. (1998) Nutritional Epidemiology 2nd ed. 1998 Oxford University Press New York, NY.
16. World Health Organization, Food and Agriculture Organization (2003) Diet, Nutrition and the Prevention of Chronic Diseases. Report of a Joint WHO/FAO Expert Consultation 2003 WHO Geneva, Switzerland.
17. European Heart Network (1998) Food, Nutrition and Cardiovascular Disease Prevention in Europe 1998 European Heart Network Brussels, Belgium.
18. Hu, F. B. & Willett, W. C. (2002) Optimal diets for prevention of coronary heart disease. J. Am. Med. Assoc. 288:2569-2578.
19. Lampe, J. (1999) Health effects of vegetables and fruits: assessing mechanisms of action in human experimental studies. Am. J. Clin. Nutr. 70(suppl.):475S-490S.
20. Dragsted, L. O., Strube, M. & Larsen, J. C. (1993) Cancer-protective factors in fruits and vegetables: biochemical and biological background. Pharmacol. Toxicol. 72(suppl. 1):116-135.
21. Powles, J., Day, N. E., Sanz, M. A. & Bingham, S. A. (1996) Protective foods in winter and spring: a key to lower vascular mortality?. Lancet 348:898-899.[Medline]
22. Rehm, J., Monteiro, M., Room, R., Gmel, G., Jernigan, D., Frick, U. & Graham, K. (2001) Steps towards constructing a global comparative risk analysis for alcohol consumption: determining indicators and empirical weights for patterns of drinking, deciding about theoretical minimum, and dealing with different consequences. Eur. Addict. Res. 7:138-147.[Medline]
23. Kauhanen, J., Kaplan, G. A., Goldberg, D. E., Salonen, R. & Salonen, J. T. (1999) Pattern of alcohol drinking and progression of atherosclerosis. Arterioscler. Thromb. Vasc. Biol. 19:3001-3006.
24. Mukamal, K. J., Conigrave, K. M., Mittleman, M. A., Camargo, C. A., Jr., Stampfer, M. J., Willett, W. C. & Rimm, E. B. (2003) Roles of drinking pattern and type of alcohol consumed in coronary heart disease in men. N. Engl. J. Med. 348:109-118.
25. Scheaffer, R. L., Mendenhall, W. & Ott, L. (1990) Elementary Survey Sampling 4th ed. 1990 PWS-Kent Publishing Company Boston, MA.
26. Irala-Estevez, J. D., Groth, M., Johansson, L., Oltersdorf, U., Prättälä, R. & Martinez-Gonzalez, M. A. (2000) A systematic review of socio-economic differences in food habits in Europe: consumption of fruit and vegetables. Eur. J. Clin. Nutr. 54:706-714.[Medline]
27. Armstrong, B. K., White, E. & Saracci, R. (1992) Principles of Exposure Measurement in Epidemiology 1992 Oxford University Press Oxford, UK.
28. EFCOSUM Group (2001) European Food Consumption Survey Method. TNO Report V3766 2001 TNO Nutrition and Food Research Zeist, The Netherlands.
29. Nelson, M. & Bingham, S. (1997) Assessment of food consumption and nutrient intake. Margetts, B. M. Nelson, M. eds. Design Concepts in Nutritional Epidemiology 2nd ed. 1997:123-169 Oxford University Press Oxford, UK. .
30. Johansson, G., Wikman, A., Ahren, A. M., Hallmans, G. & Johansson, I. (2001) Underreporting of energy intake in repeated 24-hour recalls related to gender, age, weight status, day of interview, educational level, reported food intake, smoking habits and area of living. Public Health Nutr. 4:919-927.[Medline]
31. Hebert, J. R., Clemow, L., Pbert, L., Ockene, I. S. & Ockene, J. K. (1995) Social desirability bias in dietary self-report may compromise the validity of dietary intake measures. Int. J. Epidemiol. 24:389-398.
32. Puska, P. (2000) Nutrition and mortality: the Finnish experience. Acta Cardiol. 55:213-220.[Medline]
33. Wild, C. P., Andersson, C., O’Brien, N. M., Wilson, L. & Woods, J. A. (2001) A critical evaluation of the application of biomarkers in epidemiological studies on diet and health. Br. J. Nutr. 86(suppl. 1):S37-S53.
This article has been cited by other articles:
|
|
 |
|
  C. Sirikulchayanonta, P. Pavadhgul, R. Chongsuwat, and S. Srisorrachata A preliminary study of hyperlipidemia in Bangkok school children Asia Pac J Public Health, September 1, 2006; 18(3): 15 - 19. [PDF]
|
|
|
|
 |
|
 S. L Thomas, J. G Wheeler, and A. J Hall Micronutrient intake and the risk of herpes zoster: a case-control study Int. J. Epidemiol., April 1, 2006; 35(2): 307 - 314. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Pomerleau, K. Lock, C. Knai, and M. McKee Interventions Designed to Increase Adult Fruit and Vegetable Intake Can Be Effective: A Systematic Review of the Literature J. Nutr., October 1, 2005; 135(10): 2486 - 2495. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
