Tobacco Documents Online

628 Effect of fresh fruit consumption on function and wheeze in children Thorax 1997;52:628-633 lung Derek G Cook, Iain M Carey, Peter H V~hincup, Olia Papacosta, Susanna Chirico, K Richard Bruckdorfer, Mary Walker Department of Public Health Sciences, St George's Hospital Medical School, London SW17 ORE, UK D G Cook I M Carcy Cardiovascular Research Group, Department of Primary Care and Population Sciences P H "~nincup 0 Papacos~a M Walker Department of Biochemistry and Molecular Medicine S Chifico K R Bruckdorfer Royal Free Hospital School of Medicine, London NW3 2PF, UK Correspondence to: Dr D G Cook. Received 16 December 1996 Returned to authors 13 january 1997 Revised version received 7 March 1997 Accepted for publication 24 March 1997 Abstract Background - Fresh fruit consumption and vitamin C intake have been associated with improved lung fimction in adults. Whether this is due to enhancement of lung growth, to a reduction in lung function decline, or to protection against broncho- spasm is unclear. Methods - In a cross-sectional school based survey of 2650 children aged 8-11 from 10 towns in England and Wales the main outcome measure was forced ex- piratory volume in one second (FEVt) standardised for body size and sex. Ex- posure was assessed by a food frequency questionnaire to parents and by measure- ment of plasma levels of vitamin C in a subsample of 278 children. Results - FEVt was positively associated with frequency of fresh fruit consumption. After adjustment for possible confounding variables including social class and passive smoking, those who never ate any fresh fi-uit had an estimated FEVt some 79 ml (4.3%) lower than those who ate these items more than once a day (95% CI 22 to 136 ml). The association between FEV~ and frnlt consumption was stronger in subjects with wheeze than in non-wheezers (p= 0.020 for difference in trend), though wheeze itself was not related to fresh fruit consumption. Frequency of consumption of salads and of green vegetables were both associated with FEV~ but the relationships were weaker than for fresh fi~it. Plasma vitamin C levels were unrelated to FEV~ (r= -0.01, p =0.92) or to wheeze and were only weakly related to fresh fruit con- sumption (r=0.13, p=0.055). Conclusions- Fresh fruit consumption ap- pears to have a beneficial effect on lung function in children. Further work is needed to confirm whether the effect is restricted to subjects who wheeze and to identify the specific nutrient involved. (Thorax 1997;52:628-633) corbate might protect against the lung damage caused by cigarette smokingfl However, the effects are not confined to smokers13 and in one study appeared to be stronger in subjects who had never smoked and to be present in young adults, suggesting that other biological mechanisms may be required to explain re- duced ventilatory function in subjects with a low fruit intake.2 Strachan suggested that the relationship in adults might be an indirect re- flection of an influence of diet on lung growth in childhood. In adults it has also been sug- gested that any effect may be much greater amongst asthmatic subjects) We have examined the cross-sectional re- lationship between reported frequency of fresh fruit, salad, and green vegetable consumption and lung function in 9-11 year old children from 10 British towns. In addition we have examined the interactions with wheeze, and the relationship between plasma vitamin C levels and lung function in a subsample. We have focused on fresh fruit because it is both an important source of vitamin C and may be more important in accounting for individual differences in intake, and because of the evi- dence linking infrequent fresh fruit con- sumpdon to reduced lung function in adults.2 Methods SAMPLE SELECTION The study was carried out in 10 towns in England and Wales, five with exceptionally high adult cardiovascular mortality rates and five with exceptionally low rates. Full details are published elsewhere.6 Because of the strong geographical association between mortality from cardiovascular and respiratory disease, this resulted in there being five towns with a high mortality from respiratory diseases and five with a low mortaliry.7 Wkhin each town a sample of 10 primary schools corresponding to infant schools included in an cadier study were selected. Within each school 50 children aged 9-I 1 were invited to participate. Keywords: lung function, children, fresh fruit, vitamin C, wheeze. Recent studies have reported infrequent fresh fruit consumption or low vitamin C intake to be associated with reduced lung function in adults?-3 Interest in vitamin C arose out of a belief that ~ccelerated decline in pulmonary function in smokers might be due to de- ficiencies of proteolyfic enzymes.4 This raised the possibility that antioxidants such as as- SURVEY PROCEDURES Ethical approval was obtained from all relevant. local research ethics committees. Bctwc .¢n April and November 1994 a team of four trained nurses, working in pairs, visited each town. Towns in high and low mortality areas were examined alternately. The I0 schools in each town were visited during a five-day period, each pair of nurses visiting one school for a whole day. Each nurse made approximately one quar- ter of all measurements in each town. This article is for individual use only and may not be thrther reproduced or stored electronically without ~Titten permission from the copyright holder. Unauthorized reproduction may re.suit in financial and i)thc~ pen~li~cs. (c) BRUl~Sti MED JOUI~,~AL t'UBL GROUF

I I I I I I i I I I I i 1 O1 i F~'esh fi'nir cousttmprhm and lung fiotcthm Standing height was measured to the last complete millimetre with shoes removed, weight was measured to the last complete 0.1 kg, and ethnic group was assessed on the basis of the child's appearance. Room tem- perature was measured to the nearest 0.1°C with an RS digital thermometer and therrno- couple. Passive smoke exposure was measured from salivary cotinine levels. Subjects were asked to collect saliva in the mouth and to spit into a plastic container. Samples were frozen within eight hours of collection for later assay by gas-liquid chromatography which can detect cotinine concentrations as low as 0.1 ng/ml,s SPIROMETRY Each team used one "Compact" pneumo- tachograph (Vitalograph Ltd, Buckingham, UK) with a paediatric mbuthpiece adapter. These measure air flow through a resistive mesh, on the Fteisch principle, and determine volumes by flow integration. The pneumo- tachographs were calibrated twice daily by pumping five litres through the instrument using a one litre precision syringe. There was no systematic calibration drift, but machine failure during the study led to replacement of one instrument in the fourth town. After a period of instruction and two practice attempts, each child performed three forced expiratory manoeuvres according to the methods re- commended by the American Thoracic ciety.9 A fourth manoeuvre was performed if the best test variation was greater than 10% after three blows. Tests were performed in the standing position; nose-clips were not used. All spirometric indices were corrected auto- matically to BTPS. Forced expiratory volume in one second (FEV0 and other indices were automatic'ally recorded for the "best" test as defined by the American Thoracic Society* - that is, the spirogram with the greatest sum of FEV1 and FVC. In addition, a measure of reproducibility, the b~st test variation, was re- corded - that is, the difference between FEVt + FVC for the-"best" test and "second best" test as a percentage of that for the "best" test. PARENTAL QUESTIONNAIRE A self-completion questionnaire was sent to the parents of all participants on the day of the examination and included a food frequency questionnaire, a respiratory questionnaire, and information on the child's birth weight and parental occupation. Information on birth weight was supplemented from birth record data in a small number of cases where the mother could not recall the birth weight of the child. For 10 food items parents were asked: "How often does this child eat the following foods? More than once a day; once a day; most days; one or two days a week; less than once a week; never", These categories were assigned a score from 5 (more than once a day) to 0 (never). We were specifically interested in five items: fresh fruit in summer, fresh fruit in winter, green salad in summer, green salad in 629 fruit and salads we created a score by averaging the summer and winter scores. Tests for trend were based on including this score as a con- tinuous variable in regression models. How- ever, because equal intervals of the score are unlikely to correspond to equal differences in nutrient intake, and because the functional form of any relationship between nutrient in- take and lung function is unknown, we present standardised lung functions for subiec~s cate- gorised into five groups based on the score: 0-0.5 (never), 1-1.5, 2-3, 3.5-4, 4.5-5 (> daily). To further examine the specificity of any relationships found with items of interest we also looked at relationships of FEV~ to five other food items: fish, cheese, poultry, red meat, and processed meat. Social class was determined for both parents on the basis of present or most recent oc- cupation as defined by the Office of Population Censuses and Surveys 1980. Analyses in this paper refer to the head of the household. Those few households to whom a social class could not be assigned were treated as a separate group; they were mostly single parent house- holds in which the mother had never worked regularly. A detailed respiratory symptom ques- tionnaire included the question: "Has this child had wheezing or whistling in the chest in the past 12 months?" BLOOD SAMPLING AND VITA~IIN C ASSAY Blood samples were taken from the older chil- dren throughout the study? In the last two towns (Tunbridge Wells and P, hondda) samples of blood were anticoagulated in acid-citrate- dextrose. After centrifugation a 0.5 ml sample of plasma ~vas mixed with an equal volume of metaphosphoric acid (10 % v/v) and snap frozen on dry ice within three hours of venepuncture. The samples were stored for a maximum of two months at -20°C before analysis with high pressure liquid chromatographyJ° The stability of a small number of samples was checked over a period of two months and was found not to decrease by more than 6.2% compared with fresh plasma as found in other studies.~ STATISTICAL METHODS All analyses were carried out using the SAS statistical package (SAS Institute Inc, North Carolina, USA). Multiple regressions were per- formed using the GLaM. procedure. In model 1 FEV, was regressed on age, sex, height, town (I0 levels), instrument (three levels), observer (four levels), and room temperature as well as fresh fruit consumption to produce adjusted estimates of FEV, by fresh fruit consumption categories. Tests for trend were obtained by including fresh fruit consumption as a con- tinuous variable. Model 2 included two ad- ditional covariates: an obesity index ((weight - 9)/height"v)t'- and its square because of evi- dence of non-linearity and social class (seven levels). Model 3 further adjusted for salivary winter, and green vegetables. For both fresh concentrations of cotinine and birth weight This article is for individual use only and may not be thrther reproduced or stored electronically without x~a-ittan permission from the copyright holder.

• 630 Table I Fresh fruit consumption in summer and winter in 2650 subjecu More than Once Most Once/twice Le~s than once per per days per once per day day week week Mort" than once per day 426 Once per day 3 Most days 3 Once/twice a week 0 Less than once per week 0 Never 0 238 75 15 2 2 400 106 41 4 I 12 440 182 28 0 0 3 350 75 2 0 1 2 152 15 0 1 fl 2 69 based on slightly fewer subjects. Birth weight was included because of evidence that res- piratory function is related to intrauterine growthJ~ Least squares estimates by fresh fruit consumption category were produced using the LSMEANS option. Results RESPONSE RATES AND MISSING DATA Of 4711 children invited, 3728 (79%) were screened. All analyses in this paper are re- stricted to white children aged 8-11 of whom 93% (3160/3407) had a satisfactory FEVt (in- cluding best test variation below 10%). Ques- tionnaire data on fresh fruit consumption were available on 2650 children with satisfactory FEV,. Both cotinine levels and birth weight were available on 2475 of these. FRESH FRUIT IN Vv'II~ER AND SUMMER There is a strong relationship between reported fresh fruit consumption in summer and winter, but with lower consumption in the winter (table 1). Similar results hold for salads. Our com- bined fruit score was correlated with both the salad score (Spearman correlation=0.38, p <0.001) and with green vegetable consumption (Spearman correlation = 0.27, p <0.001). FRESH FRUIT, SALAD AND GREEN VEGETABLE CONSUMPTION BY AGE, SEX, SOCIAL CLASS AND TOWN Daily fresh fruit consumption is more common in girls, exhibits the expected social class gra- 1.87 -- 1.86 Salad (p = 0.037) " . ......... = ......Green veg (p = 0.032) / 11.85.84 ~ 1.83 - ..-" /'" "'~'~ ~ 1.82 - ....'...-" < 1.81 ~..¢'~ ~ • 1.80 - 1.79 [ I I 0 (Never) 1 2 3 4 (More than Consumption score once daily) Figure I Relationship between frequency of consumption of fresh fruit, salad, and green (mode/ I). p vahtes for trend based on 2553 subjects ecqth dat6 ~iu all variables. Cook, Carey, Whincup, Papacosta, Chirico, Bruckddtfer, ct al Table 2 Fresh fruit consumption b~ age, sex, social class, and town Category No of % ~fore % Ne~.r subjects than daily (score 0-0.5) . (score 4.5-5) Sex Girls 1267 26.8 2.0 Boys 1383 23:6 4.4 Age (years) <10 580 26.4 3.3 i0 1406 25.2 3.1 I1 664 24.1 3.5 Social class I+II 903 32.1 3.4 III (non-manual) 315 21.3 3.5 III (manual) 839 24.7 3.5 IV+V 404 20.0 2.2 Missing 189 11.6 3.2 Town Low mortality: Esher 229 26.6 1.7 Chelmsford 291 22.3 3.4 Leatherhead 243 32.9 1.2 Bath 268 38.4 2.2 Tunbridge Wells 340 32.4 1.8 High mortality: Wigan 301 14.6 6.3 Port Talbot 265 21.9 3.8 Burnley 251 17.1 4.7 Rochdale 171 24.6 2.9 Rhondda 291 21.0 3.8 dient, and is markedly less common in northern and Welsh towns than in our southern towns (table 2). Similar patterns exist for both salad consumption and green vegetables, but the percentages of children consuming salads and green vegetables on a more than daily basis were much lower at 1.4% and 3.1%, re- spectively. Town and social class are thus po- tentially important confounding variables to allow for when looking at lung function and wheeze. FOOD FREQUENCY QUESTIONNAIRE AND FEV1 • FEV: adjusted for height, age, sex, and town, as well as for observer, instrument, and room temperature (model 1) shows a positive re- lationship with frequency of consumption of each of fresh fruit, salad and green vegetables (fig 1). Extreme groups with few subjects were combined with their neighbours so that all points are based on more than 100 subjects. Overall, the trend is strongest for fresh fruit, and in a regression model including all three measures the fresh fruit score remained stat- isticall'y significant (19 =0.029). A score based on all items (defined as the sum of the individual scores) was somewhat more strongly related to FEV, than was the fresh fruit score alone (test for trend, p=0.001 based on model 1). The frequency of consumption of four other dietary items exhibited no significant trends with FEVt (adiusting for model 1 confounders): fish (19=0.47), cheese (19=0.53), poultry (p= 0.21), and red meat (p=0.13). More frequent consumption of processed meat was weakly associated with reduced FEVt (p = 0.051), but adiustment for fresh fruit weakened the strength of the association (p = 0.11), while the coefficient for fresh fruit remained statistically significant (p = 0.031). The results for fresh fruit are presented in more detail in table 3. In our basic model 1 those who never ate fresh fruit (score = 0-0.5) had an estimated FEV~ some 88 ml lower than I1_ i

Fresh fi'nit consumptim~ and brag function 631 Table 3 Rdationship between mean (SE) FEV~ (I) and frequency of fruit consumption analysed by th~e models Fruit scm~ No. Model I Modal 2 lklodel 3 O~ - 2650) O~ = 2650) O~ - 2475) 0-0,5 (never) 86 1.767 (0.027) 1.766 (0.027) 1,770 (0.028) 1-1.5232 1,818 (0,017) 1.813 (0.018) 1.817 (0.018) 2-3 1054 1.840 (0.010) 1,834 (0.010) 1.839 (0.011) 3.5-4611 1.832 (0.012) 1,826 (0.012) 1.827 (0.012) 4.5--5 (more than daily) 667 1.855 (0.011) 1.845 (0.012) 1.850 (0.012) p value for trend0.0089 0.024 0.031 Model 1 adjusted for age, sex, height, town, instrument, observer, and room temperature. Model 2 as tbr model I plus obesity and social class. Model 3 as for model 2 plus eotinine and birth weight. Table 4 Prevalence of wheeze in the past.year and trends in FEVt in wheezer~ and non- wheezers by fruit scorn Fndr sco~e No.t % ¢'.'hee:ed Mean (SE) adjusted* FEV~ (litres) in last 12 0-0.5 (never) 83 25.3 1.756 (0.032) 1.763 (0.054) 1-1.5 223 13.5 1.829 (0.019)" ~ 1.724 (0.045) 2-3 1035 15.9 1.837 (0.011) 1.81l (0.020) 3.5-4 599 19.4 1.834 (0.013) 1.778 (0.024) 4.5-5 (more than daily) 656 18.0 1.838 (0.013) 1.873 (0.024) Test for trend (p) 0.63 0.27 0.0019 Test for difference in trends (p) 0.020 "1" Table based on 2596 subjects. Wheeze was missing on 54 of 2650 children. * Based on model 2. those who ate it more than once a day (score = 4.5-5) (95% CI 32 to 144ml). The results. were similar for boys and girls (data not shown). Adjustment for additional potential con- founding variables including obesity, social class, birth weight, and cotinine levels had little impact on either of the estimated effects. The difference between the high and low consumers was 79 ml for model 2 and 79 ml (95% CI 22 to 136) for model 3. While the low consumers stand out as having the lowest FEVt, there was evidence of a dose response across the other four groups, although this did not reach stat- istical significance (19=0.20). Overall, re- stricting our analysis to children whose best test variation was less o.r equal to 5% produced similar results - for e~ample, the difference between the two extrerhe groups was 97 ml (95% CI 35 to 158 ml) based on model 2. WHEEZE Wheezing or whistling in the chest in the past year was reported by 450 of the 2596 children (17.3%). There was no clear relationship be- tween wheezing and fresh fruit consumption (table 4, Z2 for trend=0.23, p=0.63) except for a higher prevalence of wheeze (25.3%) in those who never ate fresh fruit. This lack of relationship remained after adjusting for con- founding variables such as town and social class. Adjusting for wheeze thus had little effect on the relationship between FEVt values and fresh fruit consumption. However, in table 4 we see that FEV~ decreased with decreasing frequency of fresh fruit consumption more strongly in wheezers (p = 0.0019 for trend) than in non-wheezers (p=0.27 for trend). The difference in trends was statistically significant (19=0.020) and was unaffected by further ad- justment for cotinine levels and birth weight. It is noticeable that the trend amongst wheezers was particularly dependent on the high FEV~ Table 5 Mean (SD) plasma levels of vitamht C and fruit score Fruit sco~ NO. Plasma vitamin C O,nol/O 0-0.5 (never) 5 68.00 (38.06) 1-1.5 23 50.16 (13.91) 2-3 82 60.52 (22.35) 3.5-4 53 58.79 (20.34) 4.5-5 (more than daily) 65 63.80 (20.53) values amongst high consumers. Indeed, the wheezers in this group had higher mean FEVt values than the non-wheezers with the same levels of consumption, although the difference was not statistically significant. OTHER SPIROMETRIC INDICES Forced vital capacity (FVC) exhibited a ~imilar graded relationship with the fresh fruit score (p = 0.019 for trend, adjusted for model 3 con- founders), the difference between the extreme groups being 132ml (95% CI 53 to 210). Trends for 50% and 75% forced expiratory flow (FEF~0 and FEF~) were in the same dir- ection but were not ~tatistically significant (p = 0.49 a.nd p = 0.74, respectively), the differences between the extreme groups being 69 ml/s (95% CI -62 to 200) for FEF~0 and 12 m//s (95% CI -94 to 117) for FEFT~. PLASMA VITAMIN C A~D FRESH FRUIT CONSUMPTION Plasma vitamin C levels were available on 278 children in the last two towns; food frequency data were available on 228 of these and ad- equate FEV~ on 249. Table 5 shows the re- lationship between fresh fruit consumption and plasma vitamin C levels. There is a trend to- wards lower vitamin C levels with decreasing frequency of fresh fruit consumption, the vit- amin C levels being some 13.5%qower in the least frequent consumers compared with the most frequent category. However, there is con- siderable variation in the vitamin C levels within each group and the Spearman correlation co- efficient was only 0.13 (p=0.055). PLASMA VITAMIN C, WHEEZE AND LUNG FUNCTION The mean plasma vitamin" C level in 37 wheezers (64.1 ~tmol/1) was slightly higher than in 201 non-wheezers (60.1 ~tmol/1), but the difference was not significant (19=0.31, 95% CI for difference -4.7 to 12.6 i~mol/1). There was no apparent correlation between FEV~ standardised for height, age, sex, and plasma vitamin C levels, either overall (Spearman cor- relation coefficient = - 0.01, p = 0.92) or amongst the 34 wheezers (Spearman cor- relation coefficient =0.03, p = 0.88). Discussion These results support our prior hypotheses of an association between infrequent con- sumption of fresh fruit and impaired lung func- tion in children. Our findings are consistent This article is for individual use only and may not be thrther reproduced or stored electronically without ~ritten permission from the copyright holder. Unauthorized reproduction may result in financial and other penalities, it) BRITISH MED JOURJXIAL PUBL (:iROUP

632 Gook, Carey, Whfl~cup, Pap.acosta, Chirico, Bruckdo~fer, et al with all three previously published studies in adults,~-3 though the existence of unpublished negative studies is of some concern." MAGNITUDE OF EFFECT The association is unlikely to be due to chance. Errors in the measurement of lung function and the crude nature of our dietary questionnaire suggest that the true relationship between ventilatory function and specific dietary con- stituents is probably stronger than it appears from this analysis.~4 Despite the possible errors in ventilatory function and the crudity of our dietary measure, the estimated magnitude of the association found was not trivial. The difference in adjusted FEV~ between those eat- ing fresh fruit more than once a day and those who never ate it was 79 ml (95% C! 22 to 136ml), a 4.3% deficit. This is very similar to a previous study using an identical food frequency questionnaire in adults which re- stricted its analysis to non-smoking asympto- marie subjects and found a deficit in FEVI in low compared with high consumers of fruit of about 80 mlfl Our findings are also compatible with the 40 ml difference between the highest and lowest tertiles of dietary vitamin C intake reported by NHANES II.1 Where our data are not clear is on the issue of dose response. While there is a suggestion of dose response across all the groups (table 3), subjects who almost never consume fresh fruit stand out as having a markedly lower FEVv While our data are compatible with a dose response across the other groups, we lack clear evidence. BIAS AND CONFOUNDING Bias is an unlikely explanation for our findings since lung function was measured by nurses independent of parents returning the food fre- quency questionnaire. Moreover, both com- ponents were embedded in a much larger study and neither the nurses nor parents would have been aware of our interest in the specific hypo- thesis under study. A wide number of con- founding variables were controlled for in our analysis including passive and active smoking, possible birth weight influences, and social class and town. That the estimated effect of fresh fruit consumption was little hltered by ad- justment for all these variables suggests that residual confounding is unlikely. While we had no measure of total calorie intake, adjustment for body size (height and obesity) had little effect. We should be cautious of concluding that the critical nutrient is vitamin C. Fresh fruit and vegetable intake is as highly correlated with intake of other nutrients such as carotene, vitamin A, and vitamin E as it is with vitamin C.~ It is therefore interesting that the only study to compare the effect of intake of different nutrients on lung function found that both vitamin C and vitamin E intake were related to FEV,. but r_hat once vitamin C was controlled for there was no remaining association with vitamin E.~ We need also to consider that fre- quent consumption of fresh fruit is associated with frequent consumption of other items. While fresh fruit was clearly the individual item most strongly related to FEVt, both green vegetable and salad consumption showed weaker but similar trends and a score based on all these items showed the strongest re- lationship of all. Such a finding is compatible with the recent report that magnesium intake is associated with better lung function and reduced bronchial reactivity.~6 Magnesium is obtained from the diet, principally from cereals, nuts, green vegetables, and dairy produce. The frequency of consumption of other dietary items was not related to FEVt in our study. The lack of any relationship with fish consumption is interesting given recent reports.17 However, our study may have lacked power given that few children ate fish more than once a week. MODIFYING ROLE OF SYMPTOMATIC DISEASE The association does not appear to be due to symptomatic disease per se; wheeze was not strongly related to fruit consumption and ad- justment for it did not alter the relationship between FEV, and our fresh fruit score. A previous study in adults found that reduced vitamin C intake was associated with a higher prevalence of wheeze.~s There was no evidence of that in our study, except for an increased prevalence of wheeze in children who never ate fresh fruit. Nor did we find any difference" in plasma vitamin C levels between wheezers and non-wheezers. This contrasts with two previous studies which have reported lower levels of ascorbic acid in both the plasma and white cells in asthmatic subjects compared with comrol children~9 and adults.2° However, there was evidence that the positive association between frequency of fruit con- sumption and FEW was stronger in subjects who had wheezed in the past year. A recent American study found that the effect of es- timated vitamin C intake On FEV~ appeared somewhat stronger in adults with asthma and bronchitis, though not significantly so.~ Lack of statistical significance in that study could easily be due to the small number of subjects with symptoms. Several small studies have, in fact, looked for acute effects of vitamin C on airways reactivity.2~-~4 Two studies have re- ported that vitamin C inhibited decreases in FEV, on exercise or after methacholine,z~z2 Others have found no benefit in reducing the effect of a histamine challenge or on lung function.23 24 Our data suggest that the benefit of frequent consumption of fresh fruit and vegetables on FEV~ is restricted to children who wheeze. Given the lack of a relationship between wheeze and fresh fruit consumption, one interpretation is that the benefit is due to protection against bronchoconstriction in susceptible individuals which is compatible with a recent survey in adults.~-~ Such a conclusion would also explain why we find similar overall benefits in children to those seen in adults. I--Iowever, all who wheeze do not exhibit bronchial hyperreactivity

Fresh fruit consumption and lung function and the issue needs to be examined in a study where bronchial reactivity is measured before drawing conclusions. 1 l 1 ] 11 11 11 11 ]] PLASMA VITAMIN C LEVELS Food frequency questionnaires are a crude but practical way of collecting dietary information on an epidemiological scale. Our attempt to validate our food frequendy questionnaire as a measure, of vitamin C intake by measuring plasma vitamin C in a subsample was dis- appointing. However, our results are com- parable to those from a recent study in 6-10 year old American children. This study found a correlation between frequency of fresh fruit and vegetable consumption with serum vitamin C levels of only 0.29.1~ In fact, there are several reasons why plasma levels of vitamin C and frequency of fresh fruit consumption may not be highly correlated: (1) a single measure of plasma vitamin C levels may be a poor marker of long term intake; and (2) there are other important sources of vitamin C in the British diet.26 In 1994 iust under half the vitamin C in the British diet came from fresh fruit and fruit products while 40% came from vegetables, the most important being 12% from potatoes.27 In addition, there is the issue of vitamin sup- plements. We have been unable to identify any data on the prevalence of regular vitamin supplements in children in the UK. While these limitations may have clouded the relationship between frequency of fresh fruit consumption and plasma vitamin C levels, the relationship between fresh fruit consumption and lung func- tion remains and requires an explanation. We thank members of the Research Team (Sally Gassor RGN, Catherine Smart RGN, Louise Went RGN, Angela Murphy RGN, Valerie Wilson RGN) and the schools, parents and children for their cooperation. Salivary cotinine measurements were carried out at the Medical Toxicology Unit, New Cross Hospital (Dr C Feyerabend). This study was funded by The Wellcome Trust (grant no. 038976/Z/93/Z). MW was supported by the British Heart Foundation. 1 Schwartz J, Weiss ST. Relationship between dietary vitamin C intake and pulmonary function in the First Nadonal Health and Nutrition Examination Survey CNHANES I). Am ff Olin Nutr 1994;59:110-4. 2 Strachan DP, Cox BD, Erzinclioglu SW, Waiters DE, Whi- chelow MJ. Ventilatory function and winter fresh fruit consumption in a random sample of British adults. Thorax 1991 ;46:624-9. 3 Britton JR, Pavord ID, Richards K-A, Knox AJ, Wisniewski AF, Lewis SA, et al. Dietary antioxidant vitamin intake and lung function in the general population. Am ff Respir Grit Cam Med 1995;151:1383-7. 633 4 US Department of Health and Hu~nan Services. The health consequences of smokbtg: chronic obstntctlve htng disease. A report of the Surgeon General Washington D(~: United States Government Printing Office, 1984. 5 Anderson R, Theron AJ. Regulation by the antioxidants ascorate, cysteine and daspone of the increased extra- cellular and intracellular generation of reactive oxidants by activated phagocytes from cigarette smokers. Am Rev Respir Dis 1987;135:1027-32. 6 Whincup PH, Cook DG, Adshead F, Taylor S, Papacosta O, Walker M, et al. Cardiovascular risk factors in British children in towns with widely differing cardiovascular mortality rateL BMJ 1996;313:79-84. 7 Whincup PH, Cook DG, Strachan DP, Papacosta O. Time trends in respiratory symptoms in childhood over a 24 year period. Arch Dis Child 1993;68:729-34. 8 Feyerabend C, Russell MAH. A rapid gas-liquid chro- matographic method for the determination of cotinine and nicotine in biological fluids. J Pharm Pharmacol 1990;42: 450-2. 9 American Thoracic Society. Snowbird workshop on stand= ardization of spirometry. Ant Rev Respir Dis 1979;119: 831-8. 10 Jermings PE, Chirico S, Jones AF, Lunec J, Barnett AH. Vitamin C metabolites and microangiopathy in diabetes mellitus. Diabetes Res 1987;6:151-4. 11 Margolis SA, Ziegler RG, Helzlouer KJ. Ascorbic acid and dehydroascorbic acid measurement in human serum and plasma. Am -7 Clin Nutr 1991;54:13155-85. 12 Chinn S, Rona RJ, Gulliford MC, Hammond J. Weight- for-height in children aged 4-12 years. A new index compared to the normalized body mass index. Eur-7 Clin Nutr 1992;46:489-500. 13 Rona RJ, Gulliford MC, Chinn .S. Effects of prematurity and intrauterine growth on respiratory health and lung function in childhood. BM-7 1993;306:817-20. 14 Freudenheim JL, Johnson NE, Wardrop RI. Nutrient mis- classification bias in the odds ratio and loss of power in the Mantel test for trend. Int -7 Epidemiol 1989;18:232-8. 15 Byers T, Triebcr F, Gutuer E, Coates R, Sowell A, Leonard S, etal. The accuracy of parental reports of their children's intake of fruits and vegetables: validation of a food fie- quency questionnaire with serum levels of carotenoids and vitamins C, A and E. Epidemiolog3, 1993;4:350-5. 16 Brirton J, Pavord I, Richards K, Wisniewski A, Knox A, Lewis S, et aL Dietary magnesium, lung function, wheez- ing, and airway hyper-reactivity in a random adult popu- lation sample. Lancet 1994;344:357-62. 17 Schwartz J, Weiss ST. The relationship of dietary intake to level of pulmonary function in the first National Health and Nutrition Survey (NI-IAINES I). Eur RespirJ 1994; 7:1821-4. 18 Schwartz J, Weiss ST. Dietary factors and their relation to respiratory symptoms: the second national health and nutrition examination survey. An, J Epidemiol 1990;132: 67-76. 19 Aderele WR, Ette SI, Oduwoule O, Ikpeme SJ. Plasma vitamin C (ascorbic acid) levels in asthmatic children. Aft J Med Sci 1985;14:115-20. 20 Olusi SO, Ojutiku OO, Jessop WJE, Iboko MI. Plasma and white blood cell ascorbic acid concentrations in patients with bronchial asthma. Olin Chim Acta 1979;92:161-6. 21 Mohsenin % DuBois AB, Douglas JS. Effect of ascorbic acid on response to metacholine challenge in asthmatic subjects. Am Rev Respir Dis 1983;127:143-7. 22 Schacter EN, Schlessinger A. The attenuation of exercise induced bronchospasm by ascorbic acid. Ann Allergy 1982; 49:146-51. 23 Malo JL, Cartier A. Lack of acute effects of ascorbic acid on spirometry and airway responsiveness to histamine in subjects with asthma. Allergy Olin lmmunol 1986;78: 1153-8. 24 Ting S, Mansfield I.E, Yarbrough J. Effects of ascorbic acid on pulmonary functions in mild asthma..7 Asthma 1983; 20:39-42. 25 Soutar A, Seaton A, Brown K. Bronchial reactivity and dietary antioxidants. Thorax 1997;52:166-70. 26 Cade jE, Margetts BM. Nutrient sources /n the English diet: quantitative data from three English towns. Int J Epidemiol 1988;17:845-8. 27 Ministry of Agriculture Fisheries and Food. National Food • Survey 1994. London: HMSO, 1995.