Prepared by the Fifty-third meeting of the Joint FAO/WHO
    Expert Committee on Food Additives (JECFA)

    World Health Organization, Geneva, 2000
    IPCS - International Programme on Chemical Safety


    First draft prepared by Janis Baines

    Australia New Zealand Food Authority, Canberra, Australia

    Assessments of intake
         Assessments based on data on poundage (disappearance)
         Assessments based on model diets
         Assessments based on individual dietary records
    Evaluation of estimates of intake of erythrosine
         National estimates
         Estimates based on the draft General Standard for Food
    Conclusion and recommendations


         Erythrosine, a food additive used to impart a red colour to food,
    was last evaluated by the Committee at its thirty-seventh meeting
    (Annex 1, reference 94), when it established an ADI of 0-0.1 mg/kg bw.
    At its present meeting, the Committee assessed national intake
    assessments for erythrosine. This compound is proposed for use in a
    wide range of solid foods, in water-based flavoured non-alcoholic
    drinks, and in spirits and liqueurs in the draft General Standard for
    Food Additives (GSFA) being established by the Codex Committee on Food
    Additives and Contaminants. Calculations by the budget method indicate
    that the theoretical maximum level of use for erythrosine in solid
    foods is less than the maximum level of 400 mg/kg proposed by the
    draft GSFA and less than the maximum of 300 g/kg in beverages.
    Detailed intake assessments for erythrosine are therefore required.

         Information on erythrosine was received from seven member
    countries: Australia, Brazil, Canada, Japan, New Zealand, the United
    Kingdom, and the United States. A summary of the data submitted is
    given in Table 1. No countries reported use of food balance sheets or
    household economic surveys for assessing the intake of erythrosine.

        Table 1. Summary of erythrosine submissions

    Country                  Poundage   FBS/HES/        Model diets     Individual dietary
                             data       sales data                      records
    Australia-New Zealand                                               
    United Kingdom                                                     
    United States                                                     
    FBS, Food balance sheet; HES, household economic survey

    2.1  Assessments based on data on poundage (disappearance)

         Estimates of the amount of erythrosine available per capita on
    the basis of poundage (disappearance) data are given in Table 2 for
    three countries.

         The estimated intake of erythrosine was less than the ADI in all
    countries, although the estimates for Canada were 10 times higher than
    those for Japan or the USA. In Japan, the intake of erythrosine  per
     capita was 0.21 mg/day in 1997 and 0.24 mg/day in 1996 (Ishimitsu et
    al., 1997), corresponding to 3% and 4% of the ADI, assuming a mean
    body weight of 60 kg. Similar estimates of the intake of erythrosine
    available for use in the food supply were reported from the USA, with
    a mean intake of 0.22 mg/person per day and a 'pseudo-90th percentile
    intake' of 0.44 mg/person per day, representing 4% and 8% of the ADI,
    respectively, assuming a 60-kg body weight. In the Canadian
    submission, much higher intakes were reported, with a mean of 3.6
    mg/person per day, representing 60% of the ADI, assuming a 60-kg body

         Although the estimates of intake derived from poundage data tend
    to be overestimates of the actual intakes of additives by the average
    consumer, the estimates  per capita may be underestimates of the
    dietary intake of 'eaters only' of foods containing the additive.

    2.2  Assessments based on model diets

         Five countries submitted estimates of the intake of erythrosine
    based on model diets, details of which are summarized in Table 3. In
    order to interpret estimates based on model diets correctly, the
    assumptions made in constructing each model diet must be known. In the
    case of erythrosine, the estimates cannot be compared directly because
    different assumptions were made and different databases used.

        Table 2. Estimates of intake of erythrosine based on poundage data
    Country     Date     Population          Estimated       % ADIa  Comment
                                             intake (mg/kg
                                             bw per day)
    Canada      ?        Not reported        0.06             60     No other details 
    Japan       1996     126 166 000         0.0039          4       Assumed 60-kg 
                1997     126 166 000         0.0035          3       body weight
    United      1987     244 million
    States                mean per capita    0.004           4       Reported poundage 
                          90th percentile    0.008           8       data adjusted for 60% 
                                                                     response rate to 
                                                                     survey. Pseudo-90th 
                                                                     percentile intake 
                                                                     calculated by 
                                                                     multiplying mean 
                                                                     intake by 2
    a JECFA ADI, 0-0.1 mg/kg bw

        Table 3. Estimates of intake of erythrosine based on model diets 
    Country    Date       Survey                        Model                              Erythrosine   % ADIa   Assumptions
                                                                                           (mg/kg bw
                                                                                           per day)

    Brazil     1986       Potential weekly              Two-week food frequency                                   1986 survey reported 
                          intake of artificial          survey plus duplicate                                     only two hard candy 
                          food colours by               diet survey                                               products contained 
                          3-14-year-old children        Mean intake (eaters only)          0.01          10       additive out of 57 candy 
                          Sample, 242 children          High consumer intake               0.017         17       types; no other foods 
                                                                                                                  contained additive; 
                                                                                                                  assumed high consumer 
                                                                                                                  intake  2 mean intake 
                                                                                                                  for eaters; high consumer 
                                                                                                                  intake of 1986 adjusted 
                                                                                                                  other foods recently 
                                                                                                                  permitted to contain 
                                                                                                                  erythrosine not included 
                                                                                                                  in 1986 survey
                                                                                                                  (chewing-gum, breakfast 
                                                                                                                  cereals, sausage)

    Canada     NR         NR                            All persons, mean intake           0.095         95       Assumed actual use levels 
                                                                                                                  not maximum use levels
                                                                                                                  (recent permission for 
                                                                                                                  use in Filipino sausage

    Japan      1991-94    Total diet                    Average consumer                                          Wide range of permissions,
                          survey/National               1983 adults                        0.0004        0.4      except fresh food and 
                          nutrition intake survey       1991 adults                        0.0018        1.8      specified individual food
                                                        1994 adults                        0.0002        0.2      items; analysed food
                                                        1994 aged                          0.0005        0.5      additive concentrations
                                                        1992 children                      0.0014        1.4      (0 for undetectable); 
                                                                                                                  assumed 60-kg body weight 
                                                                                                                  for adults and 30 kg 
                                                                                                                  for children

    Table 3. (cont'd)
    Country    Date       Survey                        Model                              Erythrosine   % ADIa   Assumptions
                                                                                           (mg/kg bw
                                                                                           per day)

    United     1986-87    National survey:              High consumer
    Kingdom               7-day weighed                 EU permissions,                    0.01          13       Restricted permission 
                          record; adults                adult                                                     for cocktail cherries 
                          16-64 years                                                                             with maximum additive 
               1992       National survey:              EU permissions,                    0.05          52       level (EU); 97.5th 
                          7-day weighed record;         child                                                     percentile high consumption  
                          children 1.5-4.5 years                                                                  level; adjusted for  
                                                                                                                  individual body weight

    United     1982-88    14-day menu from MRCA         Long-term consumer                                        Limited number of 
    States                food frequency data            Mean, all respondents             0.00013       0.1      permissions; maximum  
                          (1982-87) plus portion         Mean, eaters only                 0.00037       0.4      additive levels; 90th
                          sizes from                                                                              percentile high 
                          Department of Agriculture                                                               consumption level (2 x 
                          (1987-88)                                                                               mean consumption for 
                          MRCA survey sample,                                                                     eaters; less than mean 
                          25 726, > 2 years                                                                       not reported because of 
                                                                                                                  low numbers); corrections 
                                                                                                                  for pre-mixes and drink 
                                                                                                                  bases; assumed 60-kg
                                                                                                                  body weight

    MRCA, Market Research Corporation of America; EU, European Union
    a  JECFA ADI, 0-0.1 mg/kg bw 

         The model diet used in the United Kingdom was constructed to
    estimate the intake of erythrosine by a high consumer, assuming that
    the additive was present at the maximum levels allowed in the Food
    Additive Directive of the European Commission, with data on 97.5th
    percentile food consumption from individual dietary records and
    adjustment for individual body weight. The model diet used in the USA
    was constructed to predict the intake of a long-term consumer. No
    details were available of how the Canadian estimate of erythrosine
    intake was derived.

         The Japanese and Brazilian model diets differed from the others
    in that the concentration of the additive derived analytically was
    used. The Brazilian model diet was constructed by integrating
    information from a two-week food frequency survey with analytical
    results from a duplicate diet survey of over 200 children aged 3-14
    years. The Japanese model diet integrated food additive levels with
    national food consumption data from national nutritional surveillance
    to obtain an estimate of the actual intake of erythrosine by the
    'average' consumer. The national nutritional survey is undertaken in
    the same year as each total diet survey. Erythrosine intake in Japan
    was first reported from a total diet survey conducted in 1983 (Ito,
    1988); data were also available for 1991 and 1994 for all adults, for
    1992 for children, and for 1993 for the aged. The exact age ranges of
    the last two population groups are not defined (Yada et al., 1995).

         In the Brazilian submission (University of Campinas, 1999), the
    intake of erythrosine was estimated from a survey conducted in 1986 of
    242 children aged 3-14 years to be 0.01 mg/kg bw per day, representing
    10% of the ADI. In this survey of 83 products of different brands and
    57 types of confectionery, only two hard candies contained
    erythrosine. More recent data indicate, however, that use of
    erythrosine has increased, and a limited number of chewing-gums,
    breakfast cereals, hard candies, and sausage products contain
    erythrosine. Extrapolation of food consumption data integrated with
    maximum use levels for these products indicates a maximum expected
    intake of erythrosine of an additional 0.007 mg/kg bw per day or a
    total of 0.017 mg/kg bw per day, representing 17% of the ADI.

         The Canadian model diet (Health Canada, 1999) showed a mean
    intake of erythrosine for the whole population of 3.9 mg/day or 0.065
    mg/kg bw per day, representing 65% of the ADI, assuming 60-kg body
    weight, and a maximum intake of 0.095 mg/kg bw per day or 95% of the
    ADI, assuming the addition of erythrosine from Filipino sausage.
    Actual levels of use were apparently used in the calculations, but
    neither the foods in which erythrosine is allowed nor the use levels
    were reported in the submission.

         The Japanese total diet survey (Japanese Food Additives
    Association, 1999) provided very low estimates of erythrosine intake.
    The intake reported in the 1991 total diet survey of adults was 0.11
    mg/day (approx. 2% of the ADI, assuming a 60-kg body weight); that in
    the 1992 survey of children was 0.042 mg/day (1.4% of the ADI,
    assuming a 30-kg body weight); that in the 1993 survey of the aged was
    0.028 mg/day (approx. 0.5% of the ADI, assuming a 60-kg body weight);
    and that in the 1994 survey of adults was 0.012 mg/day (approx. 0.2%
    of the ADI). For all age groups except children, fish and meat foods

    were the main sources of erythrosine (85-90%), with fruit, vegetables,
    and seaweed contributing a further 1-10%. For children, the main
    sources of intake were fruit, vegetables, and seaweed (65%), fish and
    meat contributing 25% and sugar products and confectionery a further
    10% to their total intake.

         The 'high consumer' model used in the United Kingdom (Ministry of
    Fisheries and Food, 1999) assumed that the maximum level in cocktail
    and candied cherries was 200 mg/kg and that in Bigarreaux cherries was
    150 mg/kg, These are the only permitted uses of erythrosine in the
    European Union, but only cocktail and candied cherries, including
    cherries used at 10% as an ingredient in cherry cakes, were reported
    to be consumed in the United Kingdom national dietary surveys of
    adults (Gregory et al., 1990) and young children (Gregory et al.,
    1995). The estimated intake of high consumers of cherries (97.5th
    percentile, consumers only) was 0.01 mg/kg bw per day for adults and
    0.05 mg/kg bw per day for children aged 1.5-4 years, reported to
    represent 13% and 52% of the ADI, respectively. No information was
    available on other foods that contain a level permitted in the GSFA
    but not a level permitted in the European Union.

         The submission from the United Kingdom also noted that
    erythrosine is widely used in pharmaceutical products in the European
    Union. A report in 1998 (European Commission, 1998) indicated that
    3639 products in Europe contained erythrosine. The maximum quantity of
    erythrosine that can be ingested with a single capsule, sugar-coated
    pill, or 1 mL of a liquid preparation has been estimated to be 0.013
    mg/kg bw per day (assuming 70 kg body weight). The ADI would be
    reached only by consuming five to seven pills or capsules or 5-7 mL of
    a liquid preparation per day, which is considered unlikely on a
    long-term basis.

         The long-term consumer model of the USA (Food & Drug
    Administration, 1999) was constructed by using data on food
    consumption from food frequency surveys conducted in 1982-88 by the
    Market Research Corporation of America and average portion sizes from
    the a three-day national food consumption survey conducted in 1997-88
    by the Department of Agriculture. Maximum additive levels were assumed
    for two models, one based on national use levels and one based on GSFA
    levels in the foods in which use of erthrosine is permitted. For most
    food groups, the maximum permitted levels of erythrosine in the USA
    were much lower than those of the GSFA, the concentrations in fruit
    preserves, jams, fruit toppings, fruit fillings, frozen soups,
    pourable salad dressings, and atole (cornmeal beverage) all being < 3
    mg/kg; however, the maximum permitted in decorative glazing was 200

         The estimates for long-term consumers of erythrosine were also
    low, with a mean for the whole population of 0.0013 mg/kg bw per day
    and a mean for eaters only of 0.0037 mg/kg bw per day, representing
    1.3% and 3.7% of the ADI, respectively. The intakes at the 90th
    percentile were actually lower than the mean intakes for eaters only
    since so few people reported eating foods in which erythrosine is
    permitted. The intake estimates are comparable to those obtained from
    US poundage data.

         Estimates of the intake of erythrosine derived from US food
    consumption data integrated with maximum GSFA levels and the range of
    foods in which use is allowed are considerably higher than those based
    on national levels of use and far exceed the ADI for both mean (1.1
    mg/kg bw per day or 1100% of the ADI) and high consumers of the
    additive (2.2 mg/kg bw or day or 2200% of the ADI). In this
    calculation, the main contributions to the mean total intake of
    erythrosine were water-based, flavoured drinks (62%), fruit juice
    concentrates (9%), bread and rolls (9%), and flavoured milk (5%).

    2.3  Assessments based on individual dietary records

         The joint submission from Australia and New Zealand
    (Australia-New Zealand Food Authority, 1999) provides estimates of the
    intake of erythrosine based on individual dietary records. The
    estimates of mean and 95th percentile intake were derived from
    estimates for individuals adjusted for individual body weight (Table
    4). The estimates of mean intake based on individual dietary records
    and the levels of use permitted in Australia and New Zealand were 0.02
    mg/kg bw per day for mean consumers (18.6% of the ADI) and 0.06 mg/kg
    bw per day for high consumers (63% of the ADI). Erythrosine is
    permitted for use only in maraschino cherries in Australia ad New
    Zealand. As only a small number of people eat this product (0.3% of
    the population), the estimate for 95th percentile consumers should be
    discounted. In the reverse budget method, up to 26 g of cherries could
    be consumed per day by males and 21 g/day by females before the ADI
    was exceeded. The submission concluded that this level of consumption
    is unlikely to be exceeded on a long-term basis. Nevertheless, since
    only a small amount of cherries need be eaten before the ADI is
    reached, the additive is restricted to use in maraschino cherries.

         Integration of intake estimates for Australia and New Zealand
    based on national food consumption data with maximum GSFA levels and
    the range of foods in which use is allowed results in considerably
    higher estimates than those based on national levels of use, which far
    exceed the ADI for both mean (1.7 mg/kg bw per day or 1700% of the
    ADI) and high consumers of the additive (5 mg/kg bw per day or 5000%
    of the ADI). In this calculation, the major contributors to mean total
    intake were bread and rolls (31%) and water-based, flavoured drinks
    (26%), with smaller contributions from ice creams and edible ices
    (7%), fruits and vegetables in sterile containers (8%, vegetables
    included in the calculations), flavoured milk (6%), and sauces (5%).


    3.1  National estimates

         Estimates of the intale of erythrosine were submitted by seven
    countries. Except in the approach based on poundage data, it was
    assumed that the maximum permitted national levels represented the
    levels at which the additive was present, except in Japan and Brazil,
    where analytical data were used. Intake estimates based on maximum
    permitted additive levels overestimate actual intake. Estimates of
    intake per capita based on poundage data would be expected to predict
    lower intakes than those based on actual consumption, such as model
    diets and individual dietary records. In general, this is the case;

    however, for erythrosine, intake estimates derived from poundage data
    in both Japan and the USA were similar to the estimates from the
    Japanese total diet survey and the US national long-term consumer
    model (< 5% of the ADI). The sources of erythrosine in Japan and the
    USA differ significantly. Japan permits use of erythrosine in a wide
    range of foods, with the exception of fresh food and other specific
    food items, but the concentrations found in the analyses were very
    low. In contrast, erythrosine is permitted in only a limited number of
    foods in the USA but at higher levels than those reported in the
    Japanese total diet survey. The maximum permitted levels of use in the
    USA are lower than those permitted in the GSFA, except in decorative

         In models of the intake of high consumers, such as those
    submitted by Brazil and the United Kingdom, the estimated intakes of
    erythrosine are higher than those in Japan and the USA (10-20% of the
    ADI), whereas that for Canada was much higher from both poundage data
    (60% of the ADI) and for the mean total intake (95% of the ADI). The
    estimates from the individual record model in Australia and New
    Zealand indicate potentially higher intakes, ranging from 19 to 63% of
    the ADI for mean and high consumers; however, since maraschino
    cherries are eaten by < 1% population, the estimate for 95th
    percentile consumer can be disregarded.

    3.2  Estimates based on the draft General Standard for Food Additives

         Intake estimates derived from the levels proposed in the draft
    GSFA are gross overestimates of the actual intake in any country,
    because the GSFA is generally compiled by adopting the highest level
    of use for any one food category submitted by Member States. The range
    of use specified in the draft GSFA is also much wider than in national

         Intake estimates based on GSFA levels and ranges of foods were
    submitted by Australia-New Zealand and the USA. The results were very
    similar, with intakes well in excess of the ADI for both mean and high
    consumers (Table 5).


         All of the national estimates of erythrosine intake were below
    the ADI. In assessing the risk of exceeding the ADI, non-food sources
    of erythrosine should also be considered, such as use in
    pharmaceutical products, which may contribute significant amounts to
    the total intake if consumed over a long period. The intake of
    erythrosine could exceed the ADI if the maximum levels in the draft
    General Standard for Food Additives are widely accepted at the
    national level; however, models based on the maximum levels of use
    proposed in the draft General Standard give overestimates of actual
    intake, because erythrosine will be used in only a limited number of
    red foods. Therefore, the Committee concluded that it is unlikely that
    long-term intake of erythrosine will exceed the ADI.

        Table 4. Estimated intake of erythrosine based on individual records 

    Country         Date    Survey                Model                      Erythrosine    % ADIa    Assumptions
                                                                             intake (mg/kg 
                                                                             bw per day)

    Australia-New   1995    National nutrition    Mean intake (all           0.0001         0.06      Maximum additive level
    Zealand                 survey; 24-h recall;  respondents)                                        for maraschino cherries
                            sample, 13 858;       Mean consumer intake       0.02           19        only; intake adjusted for
                            aged 2 to             (eaters only)                                       individual body weight
                            > 70 years            95th percentile consumer   0.06           63

    a JECFA ADI, 0-0.1 mg/kg bw

        Table 5. Summary of estimates of intake of erythrosine based on additive
             levels permitted within the draft General Standard for Food

    Country            Model                          Intake        % ADIa
                                                      (mg/kg bw 
                                                      per day)

    Australia-New      Individual records, mean       1.72          1720 
    Zealand            intake (all respondents)
                       Mean consumer intake           1.72          1730
                       (eaters only)
                       Individual records, 95th       5.01          5020
                       percentile consumer

    United States      Model diet, mean intake        1.1           1100
                       (all respondents are
                       Model diet, 90th percentile    2.2           2200

    a JECFA ADI, 0-0.1 mg/kg bw

    Australia-New Zealand Food Authority (1999) Submission to WHO.

    European Commission (1998) Opinion on toxicological data on colouring
    agents for medicinal products: Erythrosine. Adopted by the Scientific
    Committee on Medicinal Products and Medical Devices, 21 October 1998.

    Food & Drug Administration (1999) Submission to WHO.

    Health Canada (1999) Submission to WHO.

    Ishimitsu, S., et al. (1998) Production of food grade erythrosine in
    Japan: Estimated production by the official inspection of coal tar
    dyes (including dye aluminium lakes) in 1997.  Bull. Natl Inst.
     Health Sci., 116, 153-156 (in Japanese).

    Ito, Y., ed. (1988)  Daily Intakes of Food Additives in Japan:
     Determination of Food Additive Residues in Food (1976-1985), Tokyo:
    Ministry of Health and Welfare.

    Japanese Food Additives Association (1999) Submission to WHO.

    Ministry of Fisheries and Food (1999) Submission to WHO.

    University of Campinas (1999) Submission to WHO.

    Yada, T., et al. (1995) Intake level of erythrosine in Japan by market
    basket method: A study of daily intake in Japan of Group A food
    additives (additives that are not normally present as constituents in
    foodstuffs) by age group, food group and area.  Jpn. J. Food Chem.,
    2, 54-63 (in Japanese).

    See Also:
       Toxicological Abbreviations
       Erythrosine  (FAO Nutrition Meetings Report Series 46a)
       Erythrosine (WHO Food Additives Series 6)
       Erythrosine (WHO Food Additives Series 19)
       Erythrosine (WHO Food Additives Series 21)
       Erythrosine (WHO Food Additives Series 24)
       Erythrosine (WHO Food Additives Series 28)
       ERYTHROSINE (JECFA Evaluation)