Epidemiological Studies on Diet and Pulmonary Function

The methodological approaches used and the specific challenges of nutrition research should be taken into consideration when evaluating single study findings and, most importantly, their potential contribution to evidence-based recommendations. Apart from a few randomized intervention trials, most of the available evidence on the impact of diet on outcomes, such as lung function (FEV1, FVC, FEV1/FVC), symptoms, incidence, prevalence or severity of COPD, and its progression over time, largely comes from observational studies, either cross-sectional or, to a lesser extent, longitudinally in both the general population and at-risk or diseased subjects. The strength of some studies is the use of objective measures of lung function that limit the bias arising from self-reported or physician-diagnosed disease: Post-bronchodilator spirometry is the gold standard for the diagnosis of COPD, minimizing misclassification.
Assessment of dietary intake usually included a 24 h recall and food-frequency questionnaire, both with inherent limitations, including the poor measurement of usual intake due to daily variation in food intake (mostly for 24 h recall), the semiquantitative nature of the assessment, the measurement error, the variation in diet definitions, and the lack of generalizability of study findings among different populations [12]. To estimate the independent association of diet with lung function and COPD, in most studies, the confounding bias is tackled by performing the adjustment for multiple confounding factors known to influence pulmonary function or dietary behavior, including age, gender, BMI, physical activity, intake of other foods or nutrients, energy intake, educational level, and most importantly, tobacco exposure. Sex differences in susceptibility to COPD have been increasingly recognized, with evidence that women are at a greater risk of smoking-induced lung function impairment [28] and poorer health status for the same level of tobacco exposure compared to men [29], and that gender differences may also extend to different food choices [30]. Furthermore, the increased tobacco use recently registered in women likely contributes to the epidemic of COPD in women and influences interpretation of study results. Notably, smokers tend to follow an unhealthy diet compared to ex-smokers [31] and have a higher level of oxidative stress, which is targetable by diet. Moreover, a healthy diet may be associated with other beneficial lifestyles (e.g., higher level of physical activity, higher education, lower BMI, less smoking). Even after adjustment, residual confounding of dietary associations still remains possible and contributes to some inconsistencies across studies.
Many studies have focused on the effects of individual foods or nutrients in relation to respiratory outcomes. However, this information may not completely capture the overall effect of diet on respiratory health nor reflect real life conditions where foods or nutrients are eaten in multiple combinations [32]. Moreover, nutrient and food intakes are closely correlated in the diet, so it can be difficult to disentangle their independent effects. As an example, the lack of benefit of vitamin supplementation on lung function and hospitalization for COPD [33] may be explained, at least in part, by the fact that antioxidant regimens could be effective when adopted in the form of dietary patterns rather than individual nutrients. Dietary pattern analysis captures the quantities, proportions, variety, or combination of different foods and beverages in the diets and provides a framework to evaluate the health effects of the whole diet. This may increase the ability to highlight a stronger impact due to the cumulative effects of many features of the diet and to assess the interaction among synergistic components. This comprehensive approach is emphasized by prevailing dietary guidelines and has been used in several clinical settings, including CVD, cancer, and type 2 diabetes [12].
It should be acknowledged, however, that the evaluation of overall dietary patterns could mask the effects of individual foods or nutrients and disregard potential effects of foods or nutrients not included as components of the pattern. Therefore, the best option could be to complement and integrate data on eating patterns with those of individual components as much as possible in the same study population. Numerous different combinations of foods and nutrients may be potentially investigated as patterns of food intake, and approaches to rank and quantify adherence of study participants to these patterns have been developed to evaluate their association with disease risk. These approaches [12] include: (1) a priori-defined (hypothesis-driven) indices or scores designed to capture specific dietary patterns defined a priori on the basis of scientific evidence on the relation between food and nutrient intakes and health outcomes; these scores also allow measuring conformity to nutritional requirements and dietary guidelines, with the drawback of considerable variation in the composition of patterns across studies; and (2) data-driven (exploratory) statistical methods (cluster analysis, principal component and factor analysis, and reduced rank regression) to derive existing major patterns of food intake, with the limitation of being specific to the population investigated.