Hypertension is characterised as a multifactorial clinical condition in which blood pressure (BP) levels are persistently high. Currently, the condition is one of the leading causes of death and disability in the world, constituting a serious public health problem( Reference Malta, Moura and Souza 1 , 2 ). It is estimated that approximately 25 % of the world’s population is hypertensive and that the number of cases will increase by 60 % by 2025( Reference Kearney, Whelton and Reynolds 3 ). In Brazil, the Telephone-Based Surveillance of Risk and Protective Factors for Chronic Diseases study observed hypertension in 24·1 % of the adult population in 2014( 4 ).
According to the Brazilian Society of Hypertension( 2 ), genetic factors, age, sex, ethnicity, overweight and obesity, Na intake, alcohol consumption, physical inactivity and socio-economic factors are the main factors associated with increased BP levels. In the past decade, studies have revealed new evidence of the participation of dietary patterns of the population as potential risk and protective factors in the genesis of hypertension, especially the consumption of fruits and vegetables and daily meal frequency( Reference Miura, Greenland and Stamler 5 , Reference Odegaard, Jacobs and Steffen 6 ).
Adequate consumption of fruits and vegetables leads to decreasing BP levels because of their high concentration of minerals with hypotensive potential (particularly Mg and K) and low fat content( Reference Appel, Moore and Obarzanek 7 ), which contributes to the reduction of both systolic and diastolic blood pressure (SBP and DBP, respectively)( Reference Tsubota-Utsugi, Ohkubo and Kikuya 8 ). Studies associate the habit of skipping meals with increased serum cholesterol, TAG, fasting glucose and insulin concentration( Reference Timlin and Pereira 9 , Reference Leidy, Armstrong and Tang 10 ), but studies on the subject are scarce.
The present study was conducted in a low-income population with a high prevalence of food insecurity (FI)( Reference Salles-Costa, Pereira and Vasconcellos 11 ), which, in turn, contributes to a lower consumption of fruits and vegetables( Reference Popkin 12 ) when compared with higher-income populations. Thus, a dietary pattern that includes skipped meals and a low consumption of fruits and vegetables, resulting not only from the diet composition but also from difficult access to food, may be even more damaging( Reference Lang and Heasman 13 ). Studies show that FI has repercussions on the prevalence of chronic diseases related to nutrition, including diabetes mellitus, hypertension and obesity( Reference Castillo, Ramsey and Yu 14 , Reference Seligman, Laraia and Kushel 15 ). Thus, the objective of the present study was to evaluate the association of socio-demographic factors, the consumption of fruits and vegetables and the number of meals with increased BP in a population exposed to FI.
Methods
The present study is a population-based cross-sectional survey that was conducted in the municipality of Duque de Caxias (467 619 km2 area), located in the metropolitan region of Rio de Janeiro. The second district of Duque de Caxias (Campos Elíseos) was selected for the present study because it is one of the poorest districts in the municipality.
A cluster sampling design was used in the present study, with inverse sampling, considering three selection stages (census sectors, private permanent housing and individuals). The sample size consisted of 1125 households from the second district of the municipality of Duque de Caxias, based on the prevalence of extreme poverty in the municipality of 14·5 %( Reference Monteiro 16 ). Thus, a sample of at least 1000 households would ensure obtaining estimates for proportions with 5 % significance levels. In the first stage, of the total 322 district sectors, seventy-five were selected by systematic selection with probabilities proportional to the size of the geographic operational base (Base Operacional Geográfica) of 2000 of the Brazilian Institute of Geography and Statistics (Instituto Brasileiro de Geografia e Estatística (IBGE)). This method ensured representation by income, age group (children from 6 to 30 months and adolescents from 12 to 18 years of age) and sex stratum. In the second selection stage, fifteen households were randomly selected using an inverse sampling procedure( Reference Haldane 17 ). This consists of specifying a number of interviews set in advance rather than the total number of households to be sampled. The procedure allowed checking how many (n) units should be observed to obtain (k) successful interviews. In the present study, this ensured sampling a sufficient number of households with children aged between 6 and 36 months. The third selection stage included individuals from each household to be surveyed. Adults aged between 19 and 60 years, excluding bedridden individuals and pregnant/lactating women, were eligible.
The data were collected from April to December 2010 by a team of twelve interviewers, which included four registered dietitian nutritionists (RDN) who were properly trained for this function. In total, the interviewers received 40 h of training in questionnaire application, assessment of food consumption and the measurement of anthropometric measures and BP.
Study variables
Data were collected using a questionnaire with questions on socio-demographic factors and lifestyle habits, and anthropometric assessment was performed.
BP was measured using an Automatic Blood Pressure Monitor Omron® (model HEM-7113 INT; Healthcare), which met the European Society of Hypertension standards for accuracy and performance. The measurement protocol included two measurements per individual. BP measurements were evaluated by the RDN at the end of the questionnaire application and before the anthropometrics measurements (almost 30 min after the interviewers arrived at the house of each family). The adult received general information about the BP procedure, was invited to remain seated in a comfortable position and drinking and smoking was not allowed; the adult was asked to rest for 5 min before the first measure, and measurements were taken without clothing, because it might interfere with the BP cuff or constrict blood flow in the arm. The BP cuff was placed on the patient’s arm and the brachial artery was located to position the BP cuff; we chose the proper cuff size. RDN inflated the BP cuff until 30–40 mmHg above the person’s normal BP. If a difference >10 mmHg was found between the two measurements, a third measurement was performed after 20 min of rest, and the first measurement was discarded for analysis purposes. The mean of the two measurements was considered in the analyses. The cut-off points recommended by the Brazilian Society of Hypertension were used, and individuals with SBP≥140 mmHg and/or DBP≥90 mmHg were considered to have increased BP, which was analysed as a dichotomous variable (the presence or absence of increased BP).
Body weight was measured using a portable G-Tech® electronic scale (with a 150 kg capacity and an accuracy within 100 g; G Tech Scales & Systems), and height was assessed with a portable WCS® vertical anthropometer (with a 0–220 cm range and an accuracy within 1 mm; Wood Compact Stadiometer; Cardiomed). For the anthropometric measurements, the adults were informed to wear light clothes (shorts and T-shirts) and to remain barefoot during this stage. The measurements were performed by trained RDN according to the procedures described by Lohman et al. ( Reference Lohman, Roche and Martorell 18 ). BMI was estimated from the body weight and height values, and the classification proposed by the World Health Organization( 19 ) was considered when estimating adequate body weights.
To evaluate the influence of socio-demographic indicators and lifestyle habits on increased BP, a hierarchical model of determinants of increased BP was used in three explanatory levels: the basic determinants (1st level), the intermediate determinants (2nd level) and the immediate determinant (3rd level), as shown in Fig. 1. The basic determinants considered were the following factors: age( Reference Ntuli, Maimela and Alberts 20 ) (<30, 30–39, 40–49, 50–59 years); race/ethnicity( Reference Lessa 21 ) (white and black/mulatoes based on self-classification according to IBGE categories); marital status( Reference Ntuli, Maimela and Alberts 20 ) (married or single/separated/divorced); educational level( Reference Senicato and Barros 22 ) (≤9 years of study, >9 years of study); monthly household per capita income( Reference Dórea and Lotufo 23 ) (total household income divided by the number of residents who depend on this income, categorised into income thirds based on the 2010 monthly minimum wage, in dollars); economic classification of the families( Reference Piccini, Facchini and Tomasi 24 ) (according to the Socio-economic Scale of the Brazilian Association of Market Research; Associação Brasileira de Pesquisa de Mercados)( 25 ) (considering the educational level of the head of the household and the number of household appliances, vehicles, bathrooms and domestic workers); and the consumption of potable water (presence/absence of water filters in households).
Fig. 1 Hierarchial model of determinants of increased blood pressure among adults (Duque de Caxias/Rio de Janeiro, Brazil, 2010).
The intermediate determinants considered were as follows: the FI level of families, estimated by the Brazilian Household Food Insecurity Measurement Scale (Escala Brasileira de Medida Domiciliar de Insegurança Alimentar)( Reference Segall-corrêa, Marin-león and Melgar-quinonez 26 ) (categorised into food security, mild insecurity and moderate and severe insecurity); smoking( Reference Halperin, Gaziano and Sesso 27 ) (individuals who smoked up to 100 cigarettes over a lifetime); consumption of alcohol( Reference Thadhani, Camargo and Stampfer 28 ) (consumption of any beverage containing alcohol in the last year, including beer, wine, cachaça (distilled spirit made from sugar-cane juice), whisky or caipirinha (Brazilian drink made with cachaça and fruits), analysed as a dichotomous variable; consumption of fruits (consumption of fruits not in the form of juices and soft drinks) and vegetables (consumption of vegetables other than potatoes, yams and cassava or other roots, at least five times per week (yes/no)); and the number of daily meals (breakfast, lunch, snack and dinner, dividing the number of total meals into three categories: up to two, three or four meals per day). In the last level, overweight (BMI≥25 kg/m2) was considered.
Statistical analysis
For the statistical analysis, the percentage distribution and 95 % CI were estimated according to the categories of each study variable. A hierarchical model that considered variables related to the basic, intermediate and immediate determinants of increased BP was adopted. By using Poisson’s regression, univariate models were tested to obtain the prevalence ratio (PR) and its respective 95 % CI. Variables with P-values <0·20 were included in the multivariate analysis at each hierarchical level, and the models were fitted at each subsequent level, by keeping those variables in the models that were significantly associated at the 5 % significance level with the outcome of increased BP. The final hierarchical model was obtained by considering the fit of the variables at the three levels (adjusted PR). The parameters and respective 95 % CI were estimated for the expanded data, considering the cluster sampling design effect.
A database was created using CSPro 2.3( 29 ), and the data were doubly entered for quality control purposes. All of the analysis procedures were performed in Stata 13( 30 ).
All participants were informed about the ethical aspects of the study and were asked to sign the study’s consent form before the interviews. The present study was approved by the Ethics in Research Committee of the Rio de Janeiro State University (Universidade do Estado do Rio de Janeiro) (process number 73/2009).
Results
The final sample consisted of 1121 households from which 1549 adults were interviewed. Of these, 50·8 % were men and 49·2 % were women, 29·1 % exhibited increased BP and the most prevalent age group was 30–39 years (37·2 %). There was a predominance of adults who self-classified as black/mulatoes (66·4 %), married or cohabiting (74·4 %) and with 9–11 years of education (37·1 %). The predominant monthly household per capita income, considering multiples of the minimum wage, ranged from half to one monthly minimum wage for 39·8 % of the individuals evaluated, and the monthly minimum wage in 2010 was 510·00 Brazilian real. The 3rd level of economic classification (level ‘C’) was the most prevalent (70·6 %); FI was found in approximately half of the households (31·7 % mild FI and 10·4 % moderate/severe FI), and the consumption of potable water was found in more than two-thirds of the households. With regard to lifestyle habits, 19·5 % were smokers, 35·4 % of the adults (men and women) consumed alcoholic beverages, 45·2 and 32·4 % did not consume fruits and vegetables regularly, respectively, and 31 % had up to three meals per day (9 % up to two meals and 22 % up to three meals). Of the evaluated individuals, 62·5 % were overweight (Table 1).
Table 1 Prevalence ratio (%) and 95 % CI of high blood pressure, socio-demographic characteristics, lifestyle habits and overweight among adults (Duque de Caxias/Rio de Janeiro, Brazil, 2010) (Prevalence ratio and 95 % confidence intervals)
FI, food insecurity; USD, US dollars.
* The total values in each variable differ because of missing data.
† High blood pressure=individuals with systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg.
‡ Monthly minimum wage=at the USD-Brazilian real exchange rate of 1 July 2010, monthly per capita income of 228·27 USD, estimated in terciles.
§ According to the Socio-economic Scale of the Brazilian Association of Market Research( 25 ).
|| According to the Brazilian Household Food Insecurity Measurement Scale (Escala Brasileira de Medida Domiciliar de Insegurança Alimentar)( Reference Segall-corrêa, Marin-león and Melgar-quinonez 26 ).
¶ Considering smokers as individuals who smoked up to 100 cigarettes over a lifetime.
** Consumption of any beverage containing alcohol in the last year, including beer, wine, cachaça (distilled spirit made from sugar-cane juice), whisky or caipirinha (Brazilian drink made with cachaça and fruits).
†† Consumption of fruits not in the form of juices and soft drinks at least five times per week.
‡‡ Consumption of vegetables other than potatoes, yams and cassava or other roots at least five times per week.
§§ BMI≥25 kg/m2.
When assessing the relationship between the study variables and increased BP, age, education and FI were the socio-demographic factors that were significantly associated with increased BP. Among the variables related to lifestyle habits, there was a significant association of fruit consumption, daily meals and overweight with increased BP (Table 2).
Table 2 Proportion (%) and 95 % CI about the relationship among socio-demographic indicators, lifestyle habits and overweight according to high blood pressure (Duque de Caxias/Rio de Janeiro, Brazil, 2010) (Proportion (%) and 95 % confidence intervals)
FI, food insecurity.
* The total values of each variable differ because of missing data.
† High blood pressure=individuals with systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg.
‡ Monthly minimum wage=at the USD-Brazilian real exchange rate of 1 July 2010, monthly per capita income of 228·27 USD, estimated in terciles.
§ According to the Socio-economic Scale of the Brazilian Association of Market Research( 25 ).
|| According to the Brazilian Household Food Insecurity Measurement Scale (Escala Brasileira de Medida Domiciliar de Insegurança Alimentar)Reference Segall-corrêa, Marin-león and Melgar-quinonez (26 ).
¶ Considering smokers as individuals who smoked up to 100 cigarettes over a lifetime.
** Consumption of any beverage containing alcohol in the last year, including beer, wine, cachaça (distilled spirit made from sugar-cane juice), whisky or caipirinha (Brazilian drink made with cachaça and fruits).
†† Consumption of fruits not in the form of juices and soft drinks at least five times per week.
‡‡ Consumption of vegetables other than potatoes, yams and cassava or other roots at least five times per week.
§§ BMI≥25 kg/m2.
Considering the hierarchical model used to evaluate the effect of the study variables on increased BP, the results of the univariate analysis showed that, among the basic determinants, the eldest adults, those who self-classified as black and brown, those from families with moderate to severe FI, those with lower educational levels and those from economic classes D and E were associated with increased BP. Among the intermediate determinants, adults who reported consuming alcoholic beverages, not consuming fruits and having fewer meals per day (less than four meals) were associated with increased BP (Table 3). Overweight, which was an immediate determinant evaluated in the present study, was also associated with increased BP. After fitting the model, age (age group 50–59 years), low consumption of fruits in a week, fewer meals per day and overweight remained significantly associated with increased BP (Table 3).
Table 3 Prevalence ratio (PR) and 95 % CI of the association between hierarchical levels and high blood pressure (Duque de Caxias/Rio de Janeiro, Brazil, 2010) (Prevalence ratios and 95 % confidence intervals)
Ref., referent values; FI, food insecurity.
* High blood pressure=individuals with systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg.
† Monthly minimum wage=at the USD-Brazilian real exchange rate of 1 July 2010, monthly per capita income of 228·27 USD, estimated in terciles.
‡ According to the Socio-economic Scale of the Brazilian Association of Market Research( 25 ).
§ According to the Brazilian Household Food Insecurity Measurement Scale (Escala Brasileira de Medida Domiciliar de Insegurança Alimentar)( Reference Segall-corrêa, Marin-león and Melgar-quinonez 26 ).
|| Considering smokers as individuals who smoked up to 100 cigarettes over a lifetime.
¶ Consumption of any beverage containing alcohol in the last year, including beer, wine, cachaça (distilled spirit made from sugar-cane juice), whisky or caipirinha (Brazilian drink made with cachaça and fruits).
** Consumption of fruits not in the form of juices and soft drinks at least five times per week.
†† Consumption of vegetables other than potatoes, yams and cassava or other roots at least five times per week.
‡‡ BMI≥25 kg/m2.
Discussion
The present study found a prevalence of increased BP of 29·1 % in the studied population and also found that in addition to overweight age over 50 years and the consumption of fruits <5 d a week combined with fewer meals per day contributed to increased BP levels.
The association between age and BP corroborates findings from other studies that observed a direct and linear relationship between increased age and BP levels, particularly beginning at 60 years of age( Reference Chobanian, Bakris and Black 31 ). In a cross-sectional study conducted with approximately 5000 individuals between 20 and 60 years of age from the urban area of the cities of Paraná State, Brazil, it was observed that adults over 40 years of age of both sexes exhibited an increased prevalence of hypertension( Reference Ulbrich, Bertin and Neto 32 ). With regard to the relationship between fruit consumption and increased BP, the results of the final fitted model showed that among adults who consumed fruits less than five times per week, the PR of increased BP was 45 % higher compared with adults who consumed fruits more often. The mechanism by which fruit consumption leads to decreased BP is complex because of the presence of minerals such as K, Mg and Ca, which have a hypotensive effect related to vasodilation and increased natriuresis ( Reference Appel, Moore and Obarzanek 7 , Reference Binia, Jaeger and Hu 33 , Reference Griep, Stamler and Chan 34 ). The effect of diet on BP variation was evaluated by comparing a diet with a low content of fruits, vegetables and dairy products and with a fat content typical of the diet of Americans (control) with a diet rich in fruits and vegetables( Reference Appel, Moore and Obarzanek 7 ). At the end of the study, those authors observed that the diet rich in fruits and vegetables promoted a significant reduction of 2·8 mmHg in SBP (P-value<0·001) compared with the control diet. Similar results were observed by Miura et al.( Reference Miura, Greenland and Stamler 5 ), in a prospective cohort study performed during 7 years of follow-up with male workers aged between 41 and 57 years from Chicago and other cities in Illinois. According to the authors, the infrequent consumption of fruits was associated with an increase in SBP and DBP over time, regardless of age, body weight and consumption of other types of food.
With regard to possible factors associated with the lower inclusion of fruits in the diet of the evaluated adults, it should be emphasised that food choices are complex decisions influenced both by biological factors and by social, cultural and economic factors. Thus, the importance of the monthly household per capita income and food prices as a decisive factor on the consumption of fruits and vegetables should be noted. Studies conducted in developed countries indicate that fruits and vegetables increase family expenses because of their high cost and that economic restrictions favour diets with low contents of fruits and vegetables and high energy density( Reference Darmon, Ferguson and Briend 35 – Reference Drewnowski, Darmon and Ferguson 37 ). Thus, Brazilian studies show that the decrease in the price of fruits and vegetables and the increase in the minimum wage have contributed to greater amounts of fruits and vegetables in the total food purchased by families( Reference Claro and Monteiro 38 ). Although vegetables and fruits may be more expensive than some highly processed foods, the total cost of food based on fresh or minimally processed food is still lower in Brazil( 39 ).
In fact, the high price of food was mentioned by the adults interviewed in the present study as one of the reasons that contributed to the infrequent consumption of fruits and vegetables (34·2 %) among individuals with increased BP (data not shown). Moreover, in the region in which the study was conducted, fairs and markets that sell good-quality fruits and vegetables are scarce; therefore, access to healthy food is difficult, which contributes to the increased costs of diets based on fruits and vegetables. In addition, the studied population had a high prevalence of FI, which is strongly associated with poverty( Reference Salles-Costa, Pereira and Vasconcellos 11 ). Thus, it is important to encourage the consumption of fruits and vegetables along with policies to decrease their prices, such as access to foods from family farms.
The association between fewer daily meals with increased BP has been little explored in the literature. According to the results, individuals who reported consuming up to two meals per day were twice as likely to exhibit increased BP. Studies show that skipping meals is associated with being overweight( Reference Klok, Jakobsdottir and Drent 40 ), which, in the present study, was also found to be a risk factor for increased BP. Thus, a potential hypothesis for the association between fewer meals and increased BP would be that these individuals are more susceptible to overweight and therefore to increased BP levels.
The present study did not aim to evaluate specific meals but rather the number of daily meals. However, Holmbäck et al.( Reference Holmbäck, Ericson and Gullberg 41 ) suggest that, in general, a high eating frequency is associated with a healthy lifestyle. Currently, skipping breakfast is a well-studied factor and is related to diseases because the feeling of hunger and the desire to eat are classified as higher when skipping breakfast than when skipping other meals( Reference Astbury, Taylor and Macdonald 42 – Reference Levitsky and Pacanowski 44 ). Thus, new analyses that allow identifying the skipped meal will be able to contribute to the discussion about the topic explored here.
A possible limitation of the present study is the fact that it did not estimate the dietary Na intake of the evaluated adults. However, analyses of data on the Brazilian population show that the average Na intake in the country is excessive and above the recommended values. When analysing the 2008/2009 Household Budget Survey (Pesquisa de Orçamentos Familiares (POF)) by food availability, Sarno et al.( Reference Sarno, Claro and Bertazzi 45 ) found that the daily amount of Na available for consumption in Brazilian households was 4·7 g for a daily intake of 8368 kJ (2000 kcal), which is more than twice the recommended limit. This excessive amount of available Na was found in all Brazilian macro-regions and income classes. Araujo et al.( Reference Araujo, Bezerra and Barbosa 46 ) analysed data on individual food intake in Brazil (Brazilian National Food Survey (Inquérito Nacional de Alimentação)), also collected by POF (2008/2009), and found that inadequate and excessive Na intake (over 2300 mg) was reported in 89·3 % of men and 70 % of women. The analysis of isolated dietary Na intake is usually not considered in the analysis of dietary patterns because of the difficulty in accessing this information accurately, even if specific dietary survey methods are used. This challenge occurs because individuals find great difficulty in determining the amount of added Na used daily, and there is little information available on the content of Na in manufactured products. Even the method that is currently considered the gold standard to estimate Na intake has a number of limitations that compromise its use( Reference McLean 47 , Reference López Díaz-Ufano 48 ).
Another limitation was the method used to evaluate the consumption of fruits and vegetables, which was based on the frequency of consumption levels. Although many studies use more objective methods, such as 24-h records and semiquantitative frequency questionnaires, there are many studies in the literature that use the frequency of consumption in their analyses, indicating that this is a valid assessment method that also allows accessing information on consumption and its possible interrelationships with health and disease processes( Reference Martínez-González, de la Fuente-Arrillaga and López-Del-Burgo 49 , Reference Souza, Bezerra and Pereira 50 ).
A further limitation point is physical activity, which was not considered in the analysis. In fact, low levels of physical activity increase the prevalence of hypertension( Reference Turi, Codogno and Fernandes 51 ). Nevertheless, it should be noted that another study in the same population carried out by the same research group observed high prevalence of physical inactive activities in adults (70 %)( Reference Sá Silva, Sandre-Pereira and Salles-Costa 52 ). It is noteworthy that since the previous survey was concluded (2010) it has not been observed local public policy investments or infrastructure facilities, which could increase engagement leisure-time on physical activities by this population; probably the high prevalence of physical inactive activities is the same. Thus, the non-inclusion of this variable in the analysis model, as potential confounder, probably would not change the results given the strong association of low fruit intake, number of daily meals and overweight with hypertension.
In the present study, unlike what was observed by Shariff et al.( Reference Shariff, Sulaiman and Jalil 53 ), FI was not associated with increased BP. A possible explanation for this lack of association is the fact that the population had a high prevalence of FI, which may have minimised the effect of FI on BP. However, for being a homogeneous population, even when faced with poverty, a situation in which social determinants lose their magnitude, the families that managed to include the consumption of fruits and vegetables associated with a greater number of daily meals could exhibit protection factors for increased BP.
In summary, an association between increased BP and age, infrequent consumption of fruits, skipping meals and overweight was observed in the present study. The population studied was a low-income population exposed to FI that did not have access to healthy foods, in terms of the consumption of fruits and vegetables and greater frequency of meals. Therefore, the results found here reinforce the importance of encouraging and developing strategies that ensure access to healthy foods to minimise increased BP in populations with similar socio-economic conditions and with high level of FI as it was in the studied population.
Acknowledgements
The authors thank the Brazilian National Research Council (Conselho Nacional de Pesquisa (CNPq)) and Carlos Chagas Filho Research Support Foundation (Fundação Carlos Chagas de Apoio à Pesquisa do Estado do Rio de Janeiro (FAPERJ)) for support in this research.
This research was supported by the CNPq; grant number: Edital Universal 476344/2008-5) and FAPERJ; grants numbers: E-26/110·785/2010 and E-26/103·296/2011) for the research funding.
T. B. D. participated in data analysis, the manuscript concept and writing; A. F. P. collaborated in writing the discussion and in revising the manuscript; E. M. Y. collaborated in data analysis, writing the discussion and revising the manuscript; R. S.-C. conceptualised the study, coordinated the research and financial support, and contributed to the writing and revision of the manuscript.
The authors declare that there are no conflicts of interest.
