RESULTS

Section:

ChooseTop of pageAbstractMETHODSRESULTS <<DISCUSSIONReferencesCITING ARTICLES

Of the original sample of 2,633 individuals, 1,346 (51%) participated in the present study; a further 4% of the original population had died, 7% had moved from Nottingham, 23% declined or were unable to participate, and we were unable to establish contact with 15%. Three participants could not provide satisfactory lung function measure and were excluded from further analyses. The mean (± SD) age of the 1,343 participants analyzed was 56.1 (12.4) years, range 27–80; 51% were never-smokers, 38% ex-smokers, and 12% current smokers; 69% were atopic (seven people were unable to provide valid skin-prick results); and 25% reported wheezing in the past 12 months. Participation in 2000 was lower in the younger age groups, but in other respects the aforementioned characteristics of participants in 2000 were similar to those of the original 1991 study population (Table 1)

Mean vitamin C intake (± SD) increased significantly from 101.5 (39.9) mg/day in 1991 to 123.7 (50.9) mg/day in 2000, and magnesium intake decreased significantly from 384.5 (110.2) mg/day in 1991 to 329.5 (78.7) mg/day in 2000. The fall in magnesium intake was greater in men (86.2 [112.3] mg/day) than in women (24.0 [87.8] mg/day) and the increase in vitamin C smaller in men (18.4 [44.2] mg/day) than in women (25.8 [50.9] mg/day). Energy-adjusted intakes of these nutrients exhibited similar changes. Average total energy intake in our study population fell slightly over the 9 years from 2,091.0 (565.2) to 1,947.5 (470.4) kcal/day.

Cross-sectional Results

There were positive associations between FEV₁ and both vitamin C and magnesium intake in the 2000 survey (Table 3)

TABLE 3. Cross-sectional associations between fev₁ residuals and the energy-adjusted intake of vitamin c and magnesium in 2000

	Vitamin C 2000			Magnesium 2000
Potential Confounders Included in the Model	Main Effect per 100 mg	95% CI	Mean Effect per 100 mg	95% CI
Unadjusted	117.7	62.2–173.2	99.4	55.7–143.0
Adjusted for
Smoking history and pack years	76.4	22.4–130.3	59.2	16.5–101.8
BMI	118.3	62.8–173.7	98.6	45.9–142.2
Atopy	117.3	61.8–172.9	99.8	56.1–143.4
Social class	103.1	47.1–159.0	85.9	41.5–130.2
Fully adjusted	66.8	12.2–121.4	52.9	9.6–96.2
Fully adjusted and mutually adjusted*	43.6	−16.6–103.7	37.1	−10.6–84.7

^* Vitamin C and magnesium both in the model.

Definition of abbreviations: BMI = body mass index; CI = confidence interval; FEV₁ = forced expiratory volume in 1 second.

, and no significant evidence that the relationship was not linear (likelihood ratio test p = 0.92 for vitamin C, p = 0.09 for magnesium). The regression coefficients were unaffected by adjustment for body mass index or atopy, and reduced by adjustment for smoking and social class. The effects of vitamin C and magnesium on FEV₁ remained significant after adjustment for these potential confounders, such that a 100 mg/day higher intake of vitamin C and magnesium, respectively, were associated with a 66.8 (95% confidence interval [CI], 12.2–121.4) ml and 52.9 (95% CI, 9.6–96.2) ml higher FEV₁. Restriction of the analysis to participants who were not using vitamin or mineral supplements in 2000 marginally increased the magnitude of these estimates (78.8 [95% CI, 17.4–139.9] ml and 61.9 [95% CI, 14.2–109.7] ml, respectively). The shape of the relationships between vitamin C and magnesium intake in deciles and FEV₁ residuals in 1991 and 2000 are shown in Figures 1 and 2

Figure 1. Cross-sectional relationship between adjusted FEV₁ residuals and energy-adjusted vitamin C intake in 1991 and 2000. *Adjusted for age, sex, height, age squared, age–height, BMI, smoking history, pack years, social class, and graphed by the midpoint of declines of energy-adjusted vitamin C intake in mg/day.

Download Figure | Download Powerpoint

[More] [Minimize]

Figure 2. Cross-sectional relationship between adjusted FEV₁ residuals and energy-adjusted magnesium intake in 1991 and 2000. *Adjusted for age, sex, height, age squared, age–height, BMI, smoking history, pack years, social class, and graphed by the midpoint of deciles of energy-adjusted magnesium intake in mg/day.

Download Figure | Download Powerpoint

[More] [Minimize]

. The effects of these nutrients in 2000 were reduced and were not independently significant when mutually adjusted (Table 3), probably because vitamin C and magnesium intake were significantly correlated (r = 0.44). Vitamin E was associated positively with FEV₁ in univariate analysis, but was not independently significant after adjusting for smoking history. There was no relationship between lung function and vitamin A intake, and examining these relationships nutrients intakes without adjustment for energy intake did not appreciably change the magnitude or significance of the observed effects.

There was also a cross-sectional univariate relationship between magnesium intake and current wheeze in 2000, the odds ratio for the highest compared with the lowest quintile of magnesium being 0.67 (95% CI, 0.45–1.00). However, this effect was not independently significant after adjusting for potential confounders (odds ratio, 0.84; 95% CI, 0.55–1.28). Wheezing was not associated with vitamin C, A, or E intake, either before or after adjustment for confounders. There was no relationship between any of these nutrient intakes and atopy, and no evidence of interaction between any nutrient effect and age, atopy, or smoking history.

Longitudinal Results

Change in lung function between 1991 and 2000 was significantly associated with average vitamin C intake between 1991 and 2000, such that after adjusting for smoking and social class, a 100 mg higher average intake of vitamin C was related to a 50.8 (95% CI, 3.8–97.9) ml smaller reduction in FEV₁ (Table 4

TABLE 4. Longitudinal associations between change in fev₁ residuals and energy-adjusted average vitamin c intake

	Average Vitamin C
	Main Effect per 100 mg*	95% CI
Unadjusted	60.0	14.2–105.9
Adjusted for
Smoking history and pack years	50.0	3.7–96.3
BMI	60.0	14.2–106.0
Atopy	57.9	11.9–104.0
Social class	62.1	15.7–108.6
All the above factors	50.8	3.8–97.9

^* Mean effect of change in FEV₁ in milliliters over the 9 years.

For definition of abbreviations see Table 3.

and Figure 3)

Figure 3. Relationship between deciles of average energy-adjusted vitamin C intake and change in adjusted FEV₁ residuals between 1991 and 2000. FEV₁ residuals adjusted for age, sex, height, age squared, age–height, BMI, smoking history, pack years, social class, and graphed by the midpoint of deciles of energy-adjusted average vitamin C intake in mg/days.

Download Figure | Download Powerpoint

[More] [Minimize]

, equivalent to 5.6 (95% CI, 0.4–10.8) ml per year. This relationship was slightly stronger among participants not using vitamin supplements. Adjustment for change in body mass index, change in smoking status over the 9 years, or age did not affect these results. There was no evidence of interaction with smoking status, age, or atopy. There was no significant relationship between average magnesium intake and decline in FEV₁, no relationship between change in vitamin C or magnesium intake and change in lung function either before or after adjustment for confounders, and no association between either of these nutrient intakes and the incidence of wheezing. Similar analyses of the relationship between vitamin E and A and all of the aforesaid outcomes revealed no significant associations.

DISCUSSION

Section:

ChooseTop of pageAbstractMETHODSRESULTSDISCUSSION <<ReferencesCITING ARTICLES

In this study, we have confirmed that a cross-sectional relationship between vitamin C and magnesium intake and lung function, as previously reported in our population-based sample of adults (2, 16), was still present in cross-sectional analysis at a 9-year follow-up. We have also demonstrated for the first time in adults that higher amounts of vitamin C intake are associated with a lower rate of decline in FEV₁. We found no evidence of an association between longitudinal decline in FEV₁ and intake of magnesium, vitamin E, or vitamin A. In contrast to a previous longitudinal study for vitamin intake and asthma incidence (25), we found no relationship longitudinally or cross-sectionally between wheezing in the past year and any of these nutrient intakes. However, our sample size was substantially smaller than in the previous report (25), so the absence of a relationship in our study could be due to lower statistical power.

Our original study population was randomly selected from the electoral register of a Local Authority Area of Nottingham and is therefore likely to be a representative sample of the general population. Although participation in the present study was potentially biased by survival, nonmigration, and motivation to participate, our data suggest that the participants in 2000 were in fact broadly similar to the original population in terms of diet, smoking history, initial lung function, and history of respiratory disease. Participation in 2000 was lower in the younger age groups from the 1991 sample, but there are no a priori grounds to suspect that this will appreciably bias the relationship between nutrient intakes and lung function in either the cross-sectional or longitudinal analyses. A slightly lower percentage of smokers participated in the follow-up visit, though this it is unlikely to have had a major effect on our results because we adjusted for smoking in the analysis and found no interaction of any effect with smoking history. We adjusted nutrient values for total energy intake to control for any potential confounding by total energy intake and used an average of the 1991 and 2000 nutrient intakes as the more representative marker of average nutrient intake over the 9 years as in previous studies (20, 21, 26). The distribution of amounts of vitamin C and magnesium were slightly skewed to the right but any departure from normality was minor and not improved by log transformation. For simplicity, we therefore analyzed the untransformed nutrient values. We modeled change in FEV₁ using change in residuals; although we also explored the effect of modeling absolute change in FEV₁ over time with adjustment for confounding factors, both methods produced similar results, so we have used the former approach in the presentation of this paper.

In the 1991 survey, we found a strong association between magnesium and lung function and airway hyperresponsiveness, and several subsequent studies (17–19, 27), though not all (21) have since provided further evidence in support of these observations. Cross-sectionally in 2000, we again found a positive association between level of magnesium intake and lung function, though this effect appears to have applied predominantly with the lower deciles of intake (Figure 2). However, the absence of a longitudinal relationship in the present study argues against this nutrient having a major influence on decline in lung function in adult life, though it remains possible that magnesium may be important in the early development of lung function, given the previously reported strong association between magnesium intake and asthma or wheeze in a study in children (17). We did not measure airway hyperresponsiveness in the present study and are therefore unable to comment on the effect of magnesium intake in relation to this outcome in 2000.

Vitamin C has previously been linked with higher lung function in several studies using different measures of intake, including dietary recall (5–7, 9) and plasma vitamin C concentrations (3, 4, 8). Fresh fruit intake has also been associated with higher lung function (20, 28), and although vitamin C may play an important role in the association, other nutrients in fresh fruit may also be involved. Our cross-sectional results are consistent with these previous findings, though in contrast to previous studies, we did not find any evidence on this occasion that this effect may be stronger in smokers than in nonsmokers (2, 4). Our longitudinal analysis also provided evidence, for the first time, that higher average amounts of vitamin C intake were associated with a decrease in the decline of FEV₁ over a prospective 9-year time period.

To our knowledge, only two other studies have examined the effect of diet on lung function (as opposed to respiratory symptoms) in a longitudinal design. The first of these demonstrated that a reduction in fresh fruit consumption in adults was related to a greater decline in FEV₁ over a 7-year period, but that average fruit consumption was not related to the rate of decline (20). From this the authors hypothesized that rate of decline of lung function represented a potentially reversible effect. This study did not include detailed dietary data and was therefore not able to look at individual nutrient effects. In contrast, Butland and coworkers demonstrated that higher average apple consumption was associated with a smaller decline in lung function in adult men, but found no significant evidence of an effect of individual nutrients and no relationship between change in fruit consumption or nutrient intake and change in FEV₁ (21). Our observation of an effect of average vitamin C intake on FEV₁ is broadly consistent with these findings, but in our case, the absence of a relation between change in nutrient intake and decline in FEV₁ argues against an acutely reversible effect. Our data also suggest that the protective effect of vitamin C applies particularly to those in the lower range of intake (below 90 mg/day, see Figure 3). In addition, there is one study that randomized 12 smokers with chronic obstructive pulmonary disease for a month to take vitamin A or a placebo and found an improvement in FEV₁ in those subjects receiving the vitamin A supplements (29). These results are not in contrast with our results, as the majority of our population vitamin A intake reduced over the 9 years.

The relationship between atopy and diet is not clearly understood. We have previously reported a relationship between concentrations of vitamin E and both allergen skin sensitization and IgE concentrations, more strongly in relation to IgE concentrations (30), and evidence consistent with the observation has been reported from the European Respiratory Health Survey (31). Cross-sectionally in 2000, we found no relationship between vitamin E and allergen skin sensitization, but we do not have serum IgE data in the present study. However, we found no evidence that atopy was a major determinant of decline in FEV₁.

Overall our findings lend additional support to the hypothesis that a diet rich in the foods that provide vitamin C is likely to be beneficial for lung health, and by reducing the size of decline of lung function over time, are likely to reduce the risk of chronic obstructive pulmonary disease. It is important to examine how early in life the benefits of vitamin C or relevant foods operate and to determine more precisely which nutrients or food groups are likely to be involved, to guide the development of intervention studies and establish whether decline in lung function can be reduced through dietary change.