Comparison of fruit and vegetable frequency data from two Australian national surveys.

Abstract

(Nutr Diet 2004;61:88-97)

Objective: To compare the fruit and vegetable intakes reported in the 1995 National Nutrition Survey and the 1999-2000 Australian Diabetes, Obesity and Lifestyle Study.

Design: Two national cross-sectional surveys.

Subjects: 5604 and 11 041 adults aged 25 years and older.

Setting: All states and territories of Australia.

Main outcome measures: Frequency of intake of a range of fruit and vegetables, total frequency of fruit and of vegetable consumption and agreement between total frequency and responses to global questions about fruit and vegetable intakes. Comparisons between surveys are justified because the Apparent Consumption data showed little change in availability between the surveys.

Statistical analyses: The proportion eating individual fruit and vegetable items at least once per week, the population distributions of the total fruit and vegetable frequencies and responses to the global questions were calculated for each survey allowing for sampling strategies. Agreement between the total frequency and the global questions within each survey was examined using weighted kappa.

Results: The proportion reporting eating individual items at least weekly was similar (within 4%) between surveys for most commonly eaten foods. Population distributions of fruit intake derived from the total frequency and global questions were broadly similar in the two surveys. However, within each survey, the frequency sum and global question showed only moderate agreement (kappa = 0.5). The population distributions of vegetable intake derived from total frequency were similar in the two surveys after excluding items which may have been reported twice in the National Nutrition Survey. However, the distribution from the global vegetable question was different in the two surveys. The agreement between the total frequency and global question for vegetables within each survey was poor to fair (kappa = 0.2 or 0.3). Using the frequency totals, the top quintile of vegetable consumption in both surveys is six or more serves per day but only four or more serves per day by the global question.

Conclusion: In the absence of formal comparisons between the two surveys, this study indicates that many of the individual fruit and vegetable items are reported with reasonable consistency across time. However, there is enough variation that small changes across time in different surveys may not be due to true population change in intake. The global questions should not be compared to quantitative criteria until more work showing that they are valid for this purpose has been done.

Key words: fruit, vegetables, questionnaires, validity, nutritional epidemiology, surveillance

**********

Introduction

Australians are encouraged to increase their fruit and vegetable intake (1). Monitoring intakes to determine the effectiveness of health promotion messages requires surveys with comparable information over time. Different assessments were used in two recent surveys, the 1995 National Nutrition Survey (NNS) and the 1999-2000 AusDiab Survey (2,3). Although both contained a food frequency list and global questions asking about fruit and vegetable intakes, the lists and the global questions were not identical. The global question approach is attractive for ongoing monitoring because it lends itself to a range of survey situations including telephone surveys.

Ideally, two sets of tools are compared by administering both to the same individuals. Within each of the two surveys, the global question can be compared to the total frequency from the list to examine the question 'do people give consistent answers when asked for the same information in two different ways?'. However, neither the food frequency lists nor the global questions have been compared to each other. The apparent consumption data based on food production, imports and exports (4) show that there was little change in overall availability for most types of fruit and vegetables between these two surveys (Figure 1). The small increase in citrus consumption between 1997-98 and 1998-99 is equivalent to 200 g per capita per week or about one serve per week. Therefore, on the assumption that the underlying population consumption had not changed between surveys, we compared the two questionnaire sets. Under these circumstances, one would expect similar intake distributions from the lists in both surveys and similarly with the global questions.

The main purposes of this study are to compare the responses to the lists and global questions within each survey, and to compare the distributions between each survey. Based on these analyses, we highlight some problematic areas for monitoring intakes which may lead to improvements in questionnaire design. In addition, we present the frequency data for each fruit and vegetable item for each survey as these have not been published before. This information may assist those who are interested in the relative consumption of specific fruit and vegetable items.

[FIGURE 1 OMITTED]

Methods

The NNS was conducted between January 1995 and January 1996 in a sub-sample of the 1995 National Health Survey, a multistage area sample of private dwellings and a list sample of non-private dwellings with a known probability of selection for each person. At the end of the NNS interview, participants aged 12 years and older were asked to complete the food frequency questionnaire (FFQ) and return it by mail. Of the 16 087 persons aged 25 years and over selected, 9791 participated but only 7642 returned the FFQ (2).

The unquantified NNS FFQ (2) was adapted from the FFQ developed by the Anti-Cancer Council of Victoria (ACCV) (5.6) and instructs respondents to estimate average intake over the previous 12 months and to think about the separate ingredients in mixed foods. Instructions in relation to seasonal fruit (e.g. stone fruit) asked for inseason consumption if it was consumed fresh but annual consumption if it was consumed canned. The questionnaire contained 107 food types, including eight fruit items and 26 vegetable items, including legumes shown in the following tables. The global questions follow the frequency list. The structure and allowed responses for the food frequency list and global questions are shown in Figure 2.

The Australian Diabetes, Obesity and Lifestyle Study (AusDiab) was conducted between May 1999 and December 2000. All people aged 25 years and older were selected in stratified samples of urban and non-urban Collector Districts after excluding Districts with more than 10% Indigenous population, or fewer than 100 people aged 25 years or older, or which were classified as 100% rural (3). In all, 70% of adults completed the household questionnaire, and 55.3% of these, 11 247 adults, presented at the study centre.

The self-administered semi-quantified ACCV FFQ (5,6) ascertaining intake over the previous 12 months was completed at the study centre. It commences with the global questions, followed by a set of photographs allowing a choice of seven serving sizes for various food groups, followed by a list of 80 food types containing 13 fruit items and 25 vegetable items (including legumes) shown in the following tables. The structure and allowed responses for the food frequency list and global questions are shown in Figure 3.

Analysis

To describe the reported frequency of individual food items in the two surveys, we dichotomised all responses at once per week because this category was common to both FFQs. When appropriate, answers from several questions in one survey were also combined to simulate the equivalent question in the other survey.

The total consumption frequency of fruit and of vegetables was derived using the midpoint for frequency categories containing ranges (e.g. one to three per month was counted as two per month), setting the highest category in AusDiab as three per day and in NNS as six per day and assuming four weeks and 28 days per month. A second, 'reduced' total consumption frequency was calculated because examination of the food lists raised the possibility of some double reporting in the NNS (salad in sandwiches and as a side salad, stir-fry/mixed vegetables and vegetable casserole). In addition, we thought that respondents might not have considered some vegetables when answering the global question (e.g. legumes, garlic and vegetable juice) and so including them would bias the total when comparing it to the global question. Although avocado was listed with fruit in the AusDiab FFQ, we classed it as a vegetable. Similarly, we excluded juice from the 'reduced' NNS fruit total, and juice and canned fruit from the AusDiab 'reduced' total because the AusDiab global fruit question specified fresh fruit. The resulting daily total and reduced total frequencies were categorised to match the frequencies implied on the global questions, e.g. the category of two to three serves per day was interpreted as between two to four times per day.

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

Analyses were restricted to people aged 25 years and older as this age group was common to both surveys. People who had a missing response to any of the items were excluded. All analyses were done in Stata 7 (StataCorp, College Station TX, Stata 7 2001) allowing for clustering, stratification and weighting in the sampling strategy as appropriate. The weighting variable we used for NNS was 'ffqwt' which weights the results to the estimated resident population for the third quarter of 1995 and allows for under-enumeration and non-participation (7). In AusDiab, the clusters and strata were included and the sampling weights adjusted the results to the 1998 estimated resident population.

The agreement between the global questions and total frequencies within each survey was described using weighted kappa. Kappa corrects for the expected agreement due to chance. Values between -1 and 0 indicate agreement less than that predicted by chance. Values between 0 and 1 indicate increasing levels of agreement. For example, a kappa of 0.5 indicates that the agreement is 50% of the maximum possible agreement between two tools after allowing for the effects of chance (8). As the large sample sizes in both surveys mean that virtually all results are statistically significant, P-values have been generally ignored. Analyses were approved by the Human Research Ethics Committee of Territory Health Services and by the Menzies School of Health Research.

Results

There were usable data from 5604 NNS and 11 041 AusDiab participants (Table 1). The proportion reporting at least weekly consumption of fruit varied among the items that had identical names in the two surveys (Table 2). This was within 2% for citrus and bananas but there were larger differences for fruit juice, melons, pineapple and mango/pawpaw. Combining items in AusDiab to match the NNS item yielded variable results. Pome fruit were 7% higher whereas stone fruit were about 10% lower in AusDiab. The proportion reporting at least weekly consumption of many seasonal fruits (e.g. 20-27% for mangoes) seems too large to be accurate year-round frequencies in either survey.

When the lists were summed, 55.3% of the NNS sample, but 47.9% of the AusDiab sample, consumed fruit at least twice per day (Table 3). These fell to 44.7% and 42.2% respectively when the reduced total was calculated. Women had a higher consumption frequency than men. When answering the global fruit question, 52.5% and 44.7% of NNS and AusDiab respondents respectively reported consuming two or more serves/pieces per day (Table 3). Although the distributions of responses to the reduced total and global questions were similar within each survey, many of the respondents answered inconsistently as indicated by the kappas of 0.4 to 0.5. Dichotomising at two serves/times per day did not substantially improve the kappa.

A similar proportion of the population reported at least weekly consumption of many vegetables in both surveys (Table 4). Among those with identical descriptions--pumpkin, peas, green beans, spinach, cauliflower, carrots and onions--the variation was less than 2% between surveys. Data on broccoli, peppers and mushrooms were more variable. Potatoes were subdivided quite differently on the two lists but the combined potato frequency varied by only 2.1% between surveys.

When the lists were summed, the frequency distributions of vegetable consumption in the two surveys were dissimilar, for example 51% of respondents in the NNS reported consuming vegetables six or more times per day compared to 37.4% of AusDiab respondents (Table 5). However, the distributions of the reduced total vegetable consumptions were similar in the two surveys with approximately 6% of respondents in both surveys who consumed vegetables less than or equal to once per day and approximately 30% consumed vegetables six or more times per day. The intermediate categories were also similar. Women reported a higher total frequency than men.

However, the responses to the global questions were very different from each other and also from the totals from the same survey (Table 5). Not surprisingly, the kappas were low (0.2 to 0.3) and indicate poor to fair agreement. It was unexpected that the global vegetable question from AusDiab which asked about 'how many different vegetables' had slightly better agreement with the total frequency than did the NNS global question which asked about serves.

Discussion

The FFQs in both surveys had the same general structure, except for the order of the frequency categories, and a number of identically-worded items. Although some responses were very similar between the two surveys, others were importantly different. Given that the apparent consumption data indicate little change in overall availability of fruit and vegetables between the two surveys, these differences may reflect different non-response biases or different performance characteristics of the questionnaires.

The response rate was low in both surveys. Only 34.5% selected individuals provided usable FFQ data in NNS compared to 54.3% in AusDiab. NNS participants completed the FFQ and mailed it back after two long, face-to-face interviews, including a 24-hour dietary recall. In AusDiab, those who attended the centre after the household interview completed the FFQ. Weighting was used in both surveys to correct for the non-response but this can only be done using available demographic information and cannot correct for differences which were not ascertained.

The similarities and discrepancies between the two surveys have consequences for using the survey data. One potential use is to develop a score to monitor dietary diversity. Foods that were commonly eaten tended to have similar proportions reporting eating them at least weekly, whereas the results were much more variable for the less commonly eaten foods. As the latter are the foods that would tend to differentiate between individuals with more or less diverse diets there may be an important level of misclassification for diversity scores.

The pattern of agreement within and between surveys was variable. For both fruit and vegetables, the reduced totals in the two surveys were more similar in distribution than the totals. For fruit, this suggests that respondents may vary in the way that they include juice. For vegetables, the major effect was the apparent double counting of certain salad and mixed vegetable items in the NNS. The results for both groups of food indicate that people who say 'never' in response to one questioning style may report eating these foods when asked in a different way. The similarity in responses to the fruit global questions suggests that the words 'serve' and 'piece' are understood in much the same way. However, responses to the vegetable global questions indicate that 'serve' and 'kind' are not equivalent concepts. It is not possible to guess whether people eat more serves or kinds of vegetables because, for example, people may count all items in a mixed salad as 'kinds' but the salad as one serve.

Some FFQs are designed to use a global question to scale intakes because a number of studies have shown that long lists overestimate consumption frequency (6.9-11). For example, the global question is used as the true measure of frequency of a food group and the relative intakes in the frequency list is used to determine the nutrient profile of the food group. This is the method used to calculate nutrient intakes from the ACCV questionnaire (6). However not all FFQs used in Australia use a scaling item. For example the FFQ developed by the CSIRO which is used in some state-wide surveys (12) does not, and this would impact on both the reported consumption frequency and nutrient intakes in the surveyed population (10).

The discrepancy in the distribution ranges for list totals and the global questions has policy consequences. The reduced totals from both surveys indicate that nearly 50% of the population consumes fruit at least twice per day and approximately 30% consumes vegetables at least six times per day. Calculating total frequency from the lists, as we did, is how the results of food frequency questionnaires that do not have a scaling item are commonly used to calculate risks of various diseases associated with food intake in case-control and cohort studies. In both surveys, the top quintile of vegetable intake is six or more serves per day but four or more serves according to the global question. A study showing that higher intakes protected against a particular disease may come to different conclusions about the most protective level of intake depending on which tool was used. Conducting a validation study to determine which of these is 'true' is clearly important for formulating advice to the population about desirable intakes.

There have not been many validation studies done on these questionnaires. The mean consumption of fruit and vegetables across the categories of the relevant NNS global question was calculated using the 24-hour recall data in the NNS (13). Adults reporting eating one serve or less of vegetables by the global question had a mean consumption of 204 g (or nearly three serves in nutritionist's terms) vegetables in the 24-hour recall compared to 332 g (or about 4.5 serves) in those reporting eating four or more serves. However, the responses for fruit were in the opposite direction. Those reporting one serve or less had a mean consumption of 70 g (or half a serve in nutritionist's terms) compared to 375 g (2.5 serves) among those who reported eating four or more serves per day on the global question (13). The frequency list from the ACCV questionnaire used in AusDiab has been compared to multiple 24-hour recalls or records and the intakes of many nutrients are under- or overestimated by twofold (6.14), but a comparison of the global question and food groupings has not been done. These analyses should warn against interpreting the responses to the global questions in a literal fashion. The global questions clearly should not be compared to quantitative targets until much more validation and refinement work has been done.

Whether the global questions can be used for trend monitoring depends on how responsive they are to important changes in intake. There were fewer categories in the NNS global questions than the AusDiab global questions. For fruit, 89.3% of the population fell into just two categories in NNS. Each of these categories was subdivided in the AusDiab question and the results fell fairly evenly across four categories. Similarly for vegetables, three categories in NNS (which included 97.5% of responses) covered the same range as six categories in AusDiab. This means that important changes may not be easy to detect. For example, someone who increases fruit intake from once per week to once per day should not alter his or her response to the global question and yet this an important change from a health promotion viewpoint.

Comparing the results between surveys when the underlying population intake does not appear to have altered substantially (4) indicates that small changes in wording or administration leads to small changes in the distribution of responses. It is therefore worth noting that the NNS global questions were used in the 2001 National Health Survey, but were administered by an interviewer using photos of serving sizes as prompt cards (15) whereas in the NNS, the questions were self-administered with no prompts. In addition, 92% of the National Health Survey population gave usable answers. Hence it would not be possible to conclude that the small apparent increase in reported consumption of fruit and vegetables in the 2001 survey (16) reflects real increases in consumption.

In summary, our analyses show that the results of national surveys using frequency lists or global questions concerning fruit and vegetable consumption must be interpreted with caution. Small changes in wording or administration techniques may artificially create small trends. There is clearly a problem with the agreement between the lists and the global questions that indicates more validation work is needed with the global questions if they are to be used in monitoring and compared to quantitative targets.

Table 1. Number of people responding to the food frequency
questionnaires, by age group and sex, 1995 National Nutrition Survey and
1999-2000 AusDiab Survey

                        National Nutrition Survey     AusDiab
                        M            F             M            F
      Age           n     % (a)  n     % (a)  n     % (a)  n     % (a)

      25-34          374  16.2    426   15.0   584   11.8   794   13.1
      35-44          655  28.7    706   27.2  1084   21.9  1454   23.9
      45-54          601  23.4    680   21.5  1330   26.8  1531   25.2
      55-64          454  14.6    518   14.6   904   18.2  1073   17.6
      65-74          416  12.0    419   13.4   708   14.3   805   13.2
      75 and older   145   5.1    210    8.4   347    7.0   427    7.0
Total               2645   -     2959    -    4957    -    6084    -

(a) The actual numbers of respondents; analyses in all other tables are
weighted to the national population.

Table 2. Fruit included in the food frequency questionnaire and
proportion of the population aged 25 years and older reporting
consumption of each item at least once per week, by sex, 1995 National
Nutrition Survey and 1999-2000 AusDiab Survey

                                        Total       Male       Female
                                      %   95% CI  %   95% CI  %   95% CI

                                         National Nutrition Survey
Fruit juice (a)                       56  54-57   57  54-59   55  52-57
Apple or pear                         71  69-72   67  65-69   75  73-76
Orange, mandarin or grapefruit        60  59-62   57  55-59   63  61-65
Banana                                75  74-76   71  69-73   79  77-80
Peach, nectarine, plum or apricot     55  53-56   47  44-49   62  60-65
Mango or pawpaw                       27  26-29   20  19-22   34  32-36
Pineapple                             22  21-24   19  18-21   25  23-27
Grapes or berries                     42  40-43   34  32-36   49  47-51
Melon (e.g. watermelon, rockmelon,
  honeydew melon)                     37  35-38   29  27-31   45  43-47

                                               AusDiab Survey
Fruit juice                           60  57-62   62  58-63   57  55-60
Apple                                 70  68-72   66  63-68   74  72-76
Pear                                  35  32-38   29  26-32   40  37-44
Apples and pears (a)(b)               78  76-79   73  71-76   82  80-83
Oranges or other citrus fruit         59  57-61   57  55-60   60  59-62
Banana                                74  72-76   70  67-73   77  75-80
Apricot                               19  18-20   14  13-16   24  23-26
Peaches or nectarines                 29  26-31   22  20-24   36  32-39
Apricots, peaches, nectarines (a)(b)  45  42-47   37  34-39   52  50-55
Mango or pawpaw                       20  18-22   15  13-17   25  22-29
Pineapple                             18  17-20   15  14-17   21  20-23
Strawberries                          24  23-26   16  15-18   32  30-35
Watermelon, rockmelon (cantaloupe),
  honeydew etc.                       31  29-34   24  21-27   39  36-41
Tinned or frozen fruit (any kind)     25  22-28   24  21-27   26  23-29

National Nutrition Survey: 2645 men, 2959 women; AusDiab: 4957 men, 6084
women.
(a) Foods not included in the 'reduced total' frequency of consumption.
(b) Variables created by combining other food items from this survey to
allow comparison with the other survey.

Table 3. Responses for global questions and the total and reduced total
frequency from the FFQ for fruit consumption, 1995 National Nutrition
and 1999-2000 AusDiab Surveys

Frequency/day                                         All
                                         FFQ (a)  FFQ (a)    Global
                                         total %  reduced %  question %

National Nutrition Survey
  Don't eat fruit                         0        0          2.4
  [less than or equal to] 1 per day      44.7     55.3       45.1
  2-3 per day                            35.7     31.1       44.2
  4-5 per day                            12.7      8.7        7.4
  [greater than or equal to] 6 per day    7.0      4.9        1.0
Kappa (b)
  Categories as above                     0.36     0.44       -
  Dichotomised at < 2 and
  [greater than or equal to] 2
    serves/times                          0.53     0.53       -

AusDiab Survey
  Don't eat fruit                         0        0.7        2.8
  [less than or equal to] 1 per day      21.8     28.7       23.5
  1 per day                              30.3     28.4       29.0
  2 per day                              21.5     19.8       26.3
  3 per day                              12.7     10.8       12.9
  [greater than or equal to] 4 per day   13.7     11.7        5.6
Kappa (b)
  Categories as above                     0.48     0.50       -
  Dichotomised at < 2 and
  [greater than or equal to] 2
    pieces/times                          0.54     0.55       -

Frequency/day                                    Male
                                         FFQ (a)     Global
                                         reduced %   question %

National Nutrition Survey
  Don't eat fruit                         0           3.2
  [less than or equal to] 1 per day      62.6        48.6
  2-3 per day                            27.5        40.2
  4-5 per day                             6.1         6.9
  [greater than or equal to] 6 per day    3.8         1.1
Kappa (b)
  Categories as above                     0.47        -
  Dichotomised at < 2 and
  [greater than or equal to] 2
    serves/times                          0.55        -

AusDiab Survey
  Don't eat fruit                         1.0         3.5
  [less than or equal to] 1 per day      35.4        29.5
  1 per day                              28.6        28.3
  2 per day                              17.9        22.8
  3 per day                               8.9        10.0
  [greater than or equal to] 4 per day    8.2         5.9
Kappa (b)
  Categories as above                     0.53        -
  Dichotomised at < 2 and
  [greater than or equal to] 2
    pieces/times                          0.55        -

Frequency/day                                    Female
                                         FFQ (a)     Global

                                         reduced %   question %

National Nutrition Survey
  Don't eat fruit                         0           1.5
  [less than or equal to] 1 per day      48.1        41.7
  2-3 per day                            34.8        48.0
  4-5 per day                            11.3         7.9
  [greater than or equal to] 6 per day    5.9         0.9
Kappa (b)
  Categories as above                     0.41        -
  Dichotomised at < 2 and
  [greater than or equal to] 2
    serves/times                          0.50        -

AusDiab Survey
  Don't eat fruit                         0.3         2.1
  [less than or equal to] 1 per day      22.2        17.7
  1 per day                              28.2        29.7
  2 per day                              21.6        29.6
  3 per day                              12.5        15.7
  [greater than or equal to] 4 per day   15.2         5.3
Kappa (b)
  Categories as above                     0.46        -
  Dichotomised at < 2 and
  [greater than or equal to] 2
    pieces/times                          0.53        -

(a) The total frequency is the sum of all fruit items in the FFQ, the
reduced total frequency excludes fruit juice in the NNS and fruit juice
and canned fruit in AusDiab.
(b) Kappas compare the totals from the FFQ to the global question
responses in the same survey; all kappas are statistically significant.

Table 4. Vegetables included in the food frequency questionnaire and
proportion of the population aged 25 years and older reporting
consumption of each item at least once per week, by sex, 1995 National
Nutrition Survey and 1999-2000 AusDiab Survey

                                     Total        Male        Female
                                 %     95% CI  %   95% CI   %     95% CI

                                        National Nutrition Survey
Potato, boiled, mashed baked     91    90-92   90  89-92    91    90-92
Hot chips                        37    35-38   44  41-46    30    28-32
Potatoes, hot chips (a)(b)       95    94-95   95  94-96    94    93-95
Pumpkin                          63    62-65   60  57-62    66    64-68
Carrots                          84    83-85   81  79-83    86    85-88
Peas                             71    70-72   73  71-75    69    67-71
Green beans                      67    66-69   67  65-69    68    66-70
Silverbeet, spinach              27    26-29   25  23-27    29    27-31
Broccoli                         60    58-62   54  52-56    66    64-68
Cauliflower                      52    51-54   48  46-50    56    54-58
Cabbage, Brussels sprouts,
  coleslaw                       47    45-49   45  43-48    49    47-51
Zucchini, eggplant, squash       31    30-33   27  25-29    36    34-38
Onion, leeks                     77    76-79   76  74-78    79    77-81
Mushrooms                        41    40-43   40  38-42    42    40-44
Tomatoes                         86    85-87   84  82-85    88    86-89
Lettuce                          80    78-81   77  75-79    83    81-84
Celery, cucumber                 64    62-65   57  54-59    71    69-72
Capsicum                         42    41-44   40  38-43    44    42-47
Soybean, tofu (a)                 3     3-4     3   3-4.5    3     2.5-4
Baked beans (a)                  19    18-21   20  19-22    18    17-20
Other beans, lentils (a)         12    11-13   12  11-14    12.5  11-14
Salad in a sandwich (a)          73    72-75   70  68-72    76    75-78
Side salad (a)                   74    73-75   72  70-74    76    74-78
Sweet potatoes                   14    13-15   12  11-14    15    13-16
Sweetcorn, corn on the cob       31    30-33   30  28-32    32    30-35
Stir-fried, mixed
  vegetables (a)                 60    58-61   57  55-60    62    60-64
Vegetable casserole (a)          24    22-25   22  20-24    25    23-27
Vegetable juice (a)               8     7-9     8   7-9      8     7-9

                                               AusDiab Survey
Potatoes, roasted or fried
  (include hot chips)            60    58-62   64  61-68    56    54-59
Potatoes cooked without fat      79    76-81   77  74-79    81    78-84
Potato (fat & no fat) (a)(b)     93    91-94   93  91-95    92    91-93
Pumpkin                          62    58-67   57  52-62    67    63-71
Carrots                          85    82-88   82  78-85    88    86-91
Peas                             69    66-73   71  67-74    68    65-72
Green beans                      67    65-69   64  61-67    69    68-71
Spinach, silverbeet              29    27-30   25  23-27    32    30-34
Broccoli                         68    64-71   62  58-65    74    70-78
Cauliflower                      53    51-56   51  48-53    56    52-59
Cabbage, Brussels sprouts        45    42-47   43  41-45    46    43-50
Zucchini                         44    41-48   40  37-43    49    45-52
Onion, leeks                     76    73-78   72  69-74    80    77-83
Mushroom                         51    47-55   48  44-52    54    49-58
Tomato sauce, tomato paste or
  dried tomatoes                 60    56-63   63  59-66    57    53-61
Fresh or tinned tomato           70    68-73   68  65-71    73    70-75
All tomatoes (a)(b)              86    84-88   86  84-88    86    83-88
Lettuce, endive or other salad
  greens                         81    78-84   79  75-82    84    81-86
Cucumber                         57    54-60   50  46-54    63    61-66
Celery                           46    44-48   38  36-41    53    51-54
Cucumber, celery (a)(b)          74    71-76   67  64-70    80    78-82
Peppers (capsicum)               50    47-54   48  44-51     3    49-57
Soy beans, soy bean curd or
  tofu (a)                        7     5-9     7   4.5-10   7     5.5-9
Baked beans (a)                  26    24-28   26  24-29    25    23-27
Other beans (include chick
  peas, lentils etc.) (a)        13    11-15   12  10-14    14    12-15
Bean sprouts or alfalfa sprouts  18.5  17-20   16  14-18    21    19-23
Beetroot                         41    37-44   37  34-41    44    40-47
Garlic (a)                       54    50-58   47  43-52    60    55-64
Avocado                          27    23-30   22  19-26    31    26-36

National Nutrition Survey: 2645 men, 2959 women; AusDiab: 4957 men, 6084
women.
(a) Foods not included in the 'reduced total'.
(b) Variables created by combining other food items from this survey to
allow comparison with the other survey.

Table 5. Responses for global short question in the 1995 National
Nutrition Survey and the total and reduced total frequency for vegetable
consumption from the food frequency questionnaires in the 1995 National
Nutrition and 1999-2000 AusDiab surveys

Response Category                                     All
                                         FFQ (a)  FFQ (a)    Global
                                         total %  reduced %  question %

National Nutrition Survey
  Don't eat vegetables                    0        0          0.5
  [less than or equal to] 1 per day       2.2      6.5       23.0
  2-3 per day                            15.7     28.8       56.0
  4-5 per day                            30.8     35.3       18.5
  [greater than or equal to] 6 per day   51.3     29.4        2.0
Kappa (b)
  Categories as above                     0.07     0.14       -
  Dichotomised at < 4 and
  [greater than or equal to] 4
    serves/times                          0.08     0.16       -

AusDiab Survey
  < 1 per day                             1.1      1.5        1.9
  1 per day                               4.0      5.3        5.9
  2 per day                               8.9     10.5       18.5
  3 per day                              14.5     16.3       39.3
  4 per day                              17.3     19.6       22.8
  '5 per day                             16.8     15.3        8.2
  [greater than or equal to] 6 per day   37.4     31.6        3.4
Kappa (b)
  Categories as above                     0.19     0.23       -
  Dichotomised at < 4 and
  [greater than or equal to] 4
    kinds/times                           0.22     0.26       -

Response Category                                Male
                                         FFQ (a)     Global
                                         reduced %   question %

National Nutrition Survey
  Don't eat vegetables                    0           0.7
  [less than or equal to] 1 per day       8.1        25.8
  2-3 per day                            32.9        56.7
  4-5 per day                            34.9        14.9
  [greater than or equal to] 6 per day   24.1         2.0
Kappa (b)
  Categories as above                     0.15        -
  Dichotomised at < 4 and
  [greater than or equal to] 4
    serves/times                          0.15        -

AusDiab Survey
  < 1 per day                             1.8         2.4
  1 per day                               7.0         7.2
  2 per day                              13.8        22.5
  3 per day                              17.8        40.6
  4 per day                              21.5        20.0
  '5 per day                             14.3         5.2
  [greater than or equal to] 6 per day   23.9         2.1
Kappa (b)
  Categories as above                     0.23        -
  Dichotomised at < 4 and
  [greater than or equal to] 4
    kinds/times                           0.23        -

Response Category                                Female
                                         FFQ (a)     Global
                                         reduced %   question %

National Nutrition Survey
  Don't eat vegetables                    0           0.3
  [less than or equal to] 1 per day       5.0        20.2
  2-3 per day                            24.6        55.4
  4-5 per day                            35.8        22.1
  [greater than or equal to] 6 per day   34.6         2.1
Kappa (b)
  Categories as above                     0.13        -
  Dichotomised at < 4 and
  [greater than or equal to] 4
    serves/times                          0.15        -

AusDiab Survey
  < 1 per day                             1.1         1.5
  1 per day                               3.5         4.6
  2 per day                               7.3        14.6
  3 per day                              15.0        38.0
  4 per day                              17.7        25.5
  '5 per day                             16.3        11.1
  [greater than or equal to] 6 per day   39.0         4.7
Kappa (b)
  Categories as above                     0.22        -
  Dichotomised at < 4 and
  [greater than or equal to] 4
    kinds/times                           0.28        -

(a) The total frequency is the sum of all vegetable items in the FFQ,
the reduced total frequency excludes side salads, salad in sandwiches
and legumes the NNS and vegetable juice, legumes and garlic in AusDiab.
(b) All kappa are statistically significant.

Acknowledgments

The AusDiab Study was supported by: The Commonwealth Department of Health and Aged Care. Eli Lilly (Aust) Pty Ltd. Janssen--Cilag (Aust) Pty Ltd, Knoll Australia Pty Ltd, Merck Lipha S.A., Alphapharm Pty Ltd. Merck Sharp & Dohme (Aust), Roche Diagnostics, Servier Laboratories (Aust) Pty Ltd, Smith-Kline Beecham International, Pharmacia and Upjohn Pty Ltd, BioRad Laboratories Pty Ltd, HITECH Pathology Pty Ltd, the Australian Kidney Foundation, Diabetes Australia (Northern Territory), Queensland Health, South Australian Department of Human Services, Tasmanian Department of Health and Human Services, Territory Health Services, Victorian Department of Human Services and Health Department of Western Australia.

For their invaluable contribution to the field activities of AusDiab, we are enormously grateful to Annie Allman, Marita Dalton, David Dunstan, Adam Meehan, Claire Reid, Alison Stewart, Robyn Tapp and Fay Wilson.

And our special thanks goes to the local collaborating centres, including Sir Charles Gairdner Hospital (Western Australia), the Prince of Wales Hospital (New South Wales), the Menzies Centre for Population Health Research (Tasmania), the Queen Elizabeth Hospital (South Australia), the Menzies School of Health Research (Northern Territory), Queensland Health, the Monash Medical Centre (Victoria), Department of Nephrology (Victoria) and the Centre for Eye Research Australia (Victoria).

References

1. SIGNAL. Eat Well Australia. A strategic framework for public health nutrition 2000-2010 and National Aboriginal and Torres Strait Islander Nutrition Strategy and Action plan: a summary 2000-2010. Canberra: National Public Health Partnership: 2002.

2. Australian Bureau of Statistics, National Nutrition Survey Users' Guide 1995, ABS Catalogue No 4801.0. Canberra: Australian Bureau of Statistics: 1998.

3. Dunstan DW, Zimmet PZ, Welborn TA, Cameron AJ, Shaw JE, de Courten M, et al. The Australian diabetes, obesity and lifestyle study (AusDiab)-Methods and response rates. Diab Res Clin Pract 2002;57:119-29.

4. Australian Bureau of Statistics. Apparent consumption of foodstuffs 1997-98 and 1998-99. ABS Catalogue No 4306.0. Canberra: Australian Bureau of Statistics: 2000.

5. Ireland P. Jolley D. Giles G, O'Dea K, Powles J, Rutishauser I, et al. Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving and ethnically diverse cohort. Asia Pacific J Clin Nutr 1994;3:19-31.

6. Hodge A. Patterson AJ, Brown WJ, Ireland P, Giles G. The Anti-Cancer Council of Victoria FFQ: relative validity of nutrient intakes compared with weighed food records in young to middle-aged women in a study of iron supplementation. Aust N Z J Public Health 2000;24:576-83.

7. Rutishauser IHE. Getting it right: how to use the data from the 1995 National Nutrition Survey. Canberra: Commonwealth Department of Health and Aged Care: 2000. (Also http://www.sph.uq.edu.au/NUTRITION/monitoring/publications.htm).

8. Altman DG. Practical statistics for medical research. London: Chapman and Hall: 1991.

9. Serdula M, Coates R, Byers T, Mokdad A, Jewell S, Chavez N, et al. Evaluation of a brief telephone questionnaire to estimate fruit and vegetable consumption in diverse study populations. Epidemiology 1993;4:455-63.

10. Amanatidis, S, Mackerras D, Simpson JM. Comparison of two frequency questionnaires for quantifying fruit and vegetable intake. Public Health Nutr 2000;4:233-9.

11. Cade J, Thompson R, Burley V, Warm D. Development, validation and utilisation of food-frequency questionnaires--a review. Public Health Nutr 2002;5:567-87.

12. CSIRO Division of Human Nutrition. What are Australians eating? Results from the 1985 and 1990 Victorian Nutrition Surveys. Adelaide: CSIRO: 1993.

13. Rutishauser IHE, Webb K, Abraham B, Allsopp R. Evaluation of short dietary questions from the 1995 NNS. Canberra: Commonwealth Department of Health and Aged Care; 2001. (Also http://www.sph.uq.edu.au/NUTRITION/monitoring/p3.htm).

14. Ambrosini GL, van Roosbroeck SA, Mackerras D, Fritschi L, de Klerk NH, Musk AW. The reliability of ten-year dietary recall: implications for cancer research. J Nutr 2003;133:2663-8.

15. Australian Bureau of Statistics, Information Paper. 2001 National Health Survey: survey questionnaires. http://www.abs.gov.au/websitedbs/D3110122.NSF/4a255eef008309e44a255eef00061e57/2656b8 f177b47ec4ca256c5d0002bb94!OpenDocument. Accessed 20 December 2003.

16. Australian Bureau of Statistics. National Health Survey. Summary of results, Australia. ABS Catalogue No 4364.0. Canberra: Australian Bureau of Statistics: 2002.

Menzies School of Health Research, Casuarina, Northern Territory, Institute of Advanced Studies, Charles Darwin University, Northern Territory

D. Mackerras, PhD, MPH, GradDipNutrDiet, BSc, Senior Research Fellow

Menzies School of Health Research, Casuarina, Northern Territory

J. Levy, Diploma Student

International Diabetes Institute, Caulfield, Victoria 

J. Shaw, MD, MRCP, FRACP, Director of Research

P. Zimmet, AO, MD, PhD, FRACP, FRCP, FACE, FAFPHM, Director

Correspondence: D. Mackerras, Menzies School of Health Research. PO Box 41096, Casuarina NT. 0811. Email: [e-mail]