Our findings suggest that subjects who consume Street food are associated with higher prevalence rates of BMI and hypertension than the restaurant eating group. We observed a relationship between SF consumption and BMI supporting the hypothesis that SF may be a proxy indicator of other components of an unhealthy lifestyle as behaviours have been documented to aggregate [27
]. These results parallel those previously reported for fast food [4
] suggesting that frequent consumption of street food is equally an unhealthy behaviour.
Despite significant, the influence of SF consumption was rather limited. We demonstrated that body weight was a highly significant predictor of elevated SF score. We additionally examined whether SF score predicted other outcomes including body mass index. These results may indicate that a diet that includes SF consumption may lead to reduction of other possible healthier foods. Additionally, these results provide preliminary information that the relationship with weight/obesity may in fact be bi-directional. Both the number of SF specialities consumed and the frequency of their consumption may influence the SF score. Subjects in the highest quartile of SF score reported values ≥ 7. From the regression analysis (Table ), approximately 100 kg of body weight were associated with 2 points of SF score, corresponding to the habitual consumption of two specialties of SF once a month or less or, alternatively, to the consumption of one specialty of SF more than once a month. Therefore, the SF score probably reflects a tendency towards a nutritional style rather than the amount of habitual energy intake. However, further validation studies would be needed to better examine this relationship.
The relationships between SF consumption, BMI and health status may have been influenced by a selection bias that is hardly avoidable in similar studies. Subjects frequenting the SF place and restaurants respectively, may differ, mainly on socioeconomic characteristics, yet specific data other than employment was not canvassed. Observed differences (Table ) concerning employment, marital status and offspring number may also reflect such differences in outcomes independently from dietary factors. The SF score changed according to employment status so thus manager and professionals had a significantly lower consumption of street food than the employed (Figure ). Unfortunately, we were unable to further parse the data in this group to further explore this association. The significant positive correlation between the SF score and BMI even within the SF group leads us to believe sampling bias may not be socioeconomic related suggesting that diet style influenced body size probably independently from socioeconomic status.
Our study describes the consumption of local traditional SF on obesity and medical illnesses and is unique in that it addresses a critical knowledge gap. Our study, to our knowledge, is the first to describe such patterns in a European city. Such information is currently only available for modern fast food consumption [4
]. Our results further the understanding that SF has possibly contributed to the sustainability of low-income people and is analogous to modern fast food for the effects on some aspects concerning health. Furthermore, these results may be useful in targeting future specific interventions in our local geographic area.
The few published studies on SF have exclusively considered only the description of food, chemical or microbiological contaminants [13
], its limited nutritional value [19
] or have been described anecdotally [21
]. In general, street food consumption is perceived as an unhealthy behaviour [23
] which exerts a strong appeal [24
] and may be advisable to limit its consumption. It has been also described that is in some cases, subjects underwent psychological conditioning strategies for decreasing SF consumption [25
Despite the wide availability of SF in the study region, we recruited the SF group in the oldest place of SF consumption of Palermo which now functions as a SF restaurant to obtain a more homogeneous and comparable group with respect to the group RES. The manner in identifying subjects reduces, to one extent, the possible psycho-social-cultural confounding factors (for instance alcohol abuse or very low income). The frequency of different street food consumption was similar in both groups, suggesting that the Palermo's street food consumption is a diffuse cross-cultural and socio-cultural phenomenon. Our results also suggest that as one ages, SF consumption is reduced as has previously been demonstrated whereby eating habits become healthier with aging [30
This study has some important limitations. Our data is self-reported and subject to recall and measurement bias. While we relied on these self-reported measures, undoubtedly this introduces bias as to the accuracy of the values/medications reported. Structured interviews and assessments would be required, which would not be practical in this study protocol. However, the manner in which we ascertained the data was similar in both groups and emerging data does suggest that self-reported BMI may be accurate [32
]. Trust in the health-care system is known to be related to self-rated health [33
]. Recent investigations demonstrate that Italians hold trust in their national health care system [35
] and we are hopeful that less reporting bias concerning self-rated health may have occurred in this study. The manner in recruiting the two groups directly in the place of food consumption did not allow further clinical data gathering; we were also limited in time by needing to ask our questionnaire in only a few minutes. Studies have demonstrated that short surveys often have higher completion rates [36
]. We acknowledge that the demographic characteristics of each group differed, which may account for participant's response to questions and may be reflective in their food consumption.
Other factors, including time of the year may influence the extent of SF consumption. Nonetheless, the comparability of anthropometric and demographic characteristics and its relationship to SF consumption is unique to our own study. Finally, the calculated SF score was not validated but still provides interesting information. The SF score has other intrinsic limitations, for instance someone who eats street food most days of the week but only eats one item would only have a score of 2, however we may consider this as an extreme situation that is expected to have no high influence on the results given the relatively high number of subjects included in this study. We acknowledge that further validation studies would be needed to allow an accurate interpretation of what an elevated SF score means.
Due to our sampling strategy, the study cannot be generalized and our results must be referred only to local people whose nutritional habits include different degree of street food consumption. Whether our results parallel similar food practices in other countries is unknown but may be a subject for further investigation.
The prevalence of hypertension and diabetes (Additional file 1
Table B) was lower than that in the general local population [37
]. Studies have shown that those with diagnosed hypertension and diabetes tend to have better eating habits than those with undiagnosed disease [39
]. The study subjects were younger than in the general population who have been documented to have poor eating habits [42
]. Secondly, our data is self-reported and subject to recall and measurement bias.
Although we describe associations of specific foods with given medical conditions, our data needs to be confirmed with longitudinal studies. The higher prevalence of hypertension in subjects eating focaccia with milza may be linked with the fact that this food is cooked with lard, and is characterized with high contents of saturated fatty acids. These entities are associated with oxidative stress, endothelial dysfunction and cardiovascular risk [43
]. On the contrary, the main ingredient of panelle is chick-pea floor that is rich in polyunsaturated fatty acids, that of arancine is rice and potatoes are the main ingredient of crocchè. Yet without knowing the exact nutritional content of each of the foods concerned in our study, our statements are purely speculative.