Aims, design and preliminary findings of the Hellenic National Nutrition and Health Survey (HNNHS)

Invitations were sent to approximately 6000 individuals (anticipating a 70–75% response rate) in to achieve the required sample size, based on a feasibility and volunteer basis in all Greek regions, by the study’s investigators from 01.09.2013 to 31.05.2015. A total 4574 (42,5% men and 57.5% women) finally agreed to participate. The sample was distributed throughout Greece, with 47.2% of it residing in the Athens Metropolitan area, 18.5% in the region of Central Macedonia, whereas the rest was almost equally scattered throughout the country (Table 1; p for the comparisons with Official statistics by region, age group and sex > 0.7). Post-hoc assessment, accounting for large population (N > 10,000) resulted in a 92% study power, for an effect size of 1.2 (OR = 1.2). When the 15% probability of chronic disease was accounted for, the power was reduced to 84%.

Age standardization was performed using the 2011 Census as the reference population’s data to check a-posteriori that the sampled population was representative of the Greek population, as per the aim of the study (calculations are provided in the excel file provided in the Additional files 1 and 2). The population was stratified by 10 years and statistical analyses were performed. The sampled population was representative for the age groups 0–19, 20–65 and 65+, and, hence were used in the analysis. Furthermore, the prevalence of chronic diseases (surveyed) of the actual Greek population (as per census), through direct standardization, was compared to the prevalence found in the study population. The crude and adjusted odds ratios (OR) calculated by age group in total and by gender did not significantly differ, hence allowing increasing generalizability of the results.

Total HNNHS sample population included volunteers ≥6 months old that reside in Greece. Exclusion criteria included individuals (i), that did not speak Greek, (ii) women who were at that time breastfeeding or pregnant, (iii) members of the armed forces (including those that are currently undergoing their compulsory military service), (iv) individuals that reside in institutions (e.g. nursing homes, rehabilitation centers, hospice centers, psychiatric institutions, prisons, monasteries), (v) those that were unable to provide informed consent due to any cause (e.g., mental impairment, psychiatric condition, drug abuse, vision or hearing loss) unless a first degree relative was able to assist in the process.

Information was collected via a series of previously validated questionnaires, from the entire population sampled (details given in “Questionnaires in brief” section). All of the questionnaire types used in HNNHS are provided in supplementary tables, along with their validation references in Additional file 1: Tables S1-S3. Additional references are listed for those questionnaires that are not relevant in this study’s results.

Clinical examinations were performed on a subsample. More specifically, an initial interview took place at the volunteer’s house, with the use of a specially designed computer software (i.e. Computer Assisted Personal Interview (CAPI)), to minimize response biases and misclassification (minimize volunteer burden and maximize reliability of collected data). The list of questionnaires applied can be seen in the Additional file 1: Tables S1 and S2. In summary, the interviewing process included data on (i) demographics, (ii) quality of life (QoL), (iii) medical history (i.e. chronic & autoimmune diseases, depression, anxiety), (iv) breastfeeding, (v) vitamin and subscribed drug intake, (vi) memory impairment, (vii) eating habits, (viii) alcohol intake, (ix) smoking habits, (x) physical activity, (xi) sleeping habits, (xii) overall patient health, and (xiii) effects of economic crisis. The questionnaires were chosen according to the volunteer’s age, as designated by the study’s protocol (Additional file 1: Tables S1-S3).

A detailed 24-h dietary recall was obtained during this process. The volunteers were also interviewed for a second 24-h dietary recall via telephone 8–20 days after the first interview, selecting a different day, and non-consecutive, as specified by HNNHS study-protocol. Specific questionnaire structure and validated food atlases for food quantification were used depending on volunteer’s age (≥1.5–4 years old, ≥4- < 10 years old, ≥10- < 12 years old and ≥ 12 years old) in order to maximize response accuracy. More specifically, dietary intake data were collected using two automated multiple-pass 24-h dietary recalls and a Food Propensity Questionnaire (FPQ). To harmonize data collection, we based our food classification and description system on FOOdEx2 developed by EFSA [8], based on volunteers age (< 2 years old and ≥ 2 years old). Main differences between the two versions was the food list, (was shorter for the < 2 year old’s), as well as the frequency response section. The latter referred to the frequency of food intake over the last 30 days for volunteers < 2 years old, or to the past year for those ≥2 years old. Both FPQs were developed based on the Hellenic, European and International guidelines. Overall, the methods of dietary assessment were chosen as per EFSA recommendations for the harmonization of data across countries member states of the European Union [8]. Data on eating patterns and behaviors were also collected (timing of food intake, number of meals, activities performed during food consumption, place of consumption, and others) to account for their effects on individuals weight status as studies support [9‐11]. The Nutrition Data System for Research (NDSR) (developed by the University of Minnesota) was used for nutrient analysis.

At the end of the interview, volunteers were provided with a list of questionnaires (hard copy) with specific instructions, to self-complete, based on the volunteer’s age and their primary response to disease state during the interviewing process (Additional file 1: Table S2). These were to be fulfilled within a specific time period, to further reduce volunteer burden (time related) and to decrease interviewer and response bias because of the nature of the nature of the questionnaires (sensitive personal information). These questionnaires included (i) qualitative FPQ (asked to be completed by all volunteers, as explained above), (ii) perceived stress scale, (iii) perception of health control, (iii) eating behavior (iv) chronic disease specific information (onset, treatment, medical follow ups, and others), (v) pregnancy and infantile information (i.e., smoking during pregnancy, number of children, weight gain per pregnancy, infant’s birth weight/length, breastfeeding (type & duration), and others), (vi) environmental exposure, (vii) social readjustment factors due to the economic crisis, (viii) asthma related information, and (vi) gastrointestinal disorders (the Greek version of Rome III FGID questionnaires for both children and adults was completed).

Interview based questionnaires and those to be self-completed were addressed to volunteers ≥12 years old. Questionnaires related to volunteers, less than 12 years old, were addressed to his/her parent or primary guardian.

In the case of volunteers being unable to self-respond (i.e., with inhibiting health complications, adolescents with lack of knowledge in specific questions) a parent/guardian was asked to assist in the interview. The economic crisis questionnaire was answered only by one adult member per household. Information on primary respondent, or on potential help received during the process was recorded (“interviewee assistant”). A small list of questionnaires where exempt from this procedure (where the main respondent has to be the volunteer himself), due to the nature of the related questions. These included questions on (i) memory impairment, (ii) screen time and alcohol use, (≥12 years - < 18 years), (iv) smoking habits (≥12 years - < 18 years) and (v) patient health questionnaire.

Completed questionnaires were handed to the participants nearest mobile unit or were given to the experienced field investigator (who performed their initial interview), when completed. To achieve a maximum response rate, the study’s trained personnel performed kind reminders via phone calls. A total of 3180 volunteers (2682 adults and 498 children and adolescents) completed all questionnaires (67% in total; 71% for adults and 62.6% for children & adolescents). Field investigators completed a quality control check-list upon checking the completed questionnaires.

Blood samples were taken from a sub-sample of the population. More specifically, all participants were invited to provide blood samples for biochemical – hematological evaluation. Of them, 1197 (26.2% of total population; 28.7% of adult population) agreed; no age distribution differences were found between the total population and those who provided blood sample (p = 0.677). Each of these individuals visited one of the 5 mobile units where medical and anthropometrics were completed (please see Additional file 1: Table S3). All samples were collected in the morning, between 8:00 and 10:00 am, upon having fasted for at least 10 h. To assure compliance all individuals were asked if they had fasted and when their last meal was.

Experienced field investigators were from various scientific fields (dietitians, physicians, sociologists as well as dietetic and medical students), and received specialized training on the HNNHS fieldwork protocol. These specialists were involved in the development, methodology and application of study questionnaires and protocol procedure attainment was assessed with quality control testing, during field-investigation.

The study was approved by the Ethics Committee of the Department of Food Science and Human Nutrition of the Agricultural University of Athens. It was also approved by Hellenic Data Protection Authority (HDPA). All members of the staff signed confidentiality agreements. Adult volunteers were asked to sign the consent form. For minors < 13 years of age the parent or primary guardian signed the form and for volunteers between 13 and 18 years of age the consent form was asked to be signed by both (parent/ guardian and volunteer).

All questionnaires used in HNNHS, were derived based on a priori knowledge and from components of previously validated questionnaires. For this process. The outcome of interest and previous work performed in the Greek population were also considered.

For demographic characteristics (marital status, education, health insurance, employment, income and changes in employment and/or income during the economic crisis) components from NHANES [12], Behavioral Risk Factor Surveillance System (BRFSS) study [13] and NDNS [14], questionnaires were used.

The Quality of Life (QoL) questionnaire included components of (i) QoL and chronic pain components of the Healthy Days Module developed by the Center for Disease Control (CDC) [15], (ii) questions with regards to self-reported height, weight and oral health, from the Health Survey for England and the Activity Limitations Module (also CDC developed) [16].

Two questionnaires were developed for alcohol consumption; one for minors and the second for adults. For minors (≥12 years old and < 18 years old) the questionnaire was developed based on questions from the Youth Risk Behavior Survey [17], the European School Survey Project and other Drugs [18] and the Global School-based Student Health Survey (GSHS) [19]. For the adult questionnaire data from NHANES study [12], BRFSS [13], Arkansas Cardiovascular Health Examination Survey (ARCHES) [20] and Recommended Alcohol Questions by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) [21] were used. Volunteers were classified as alcohol or non-alcohol consumers, based on their intake over the past 30 days. Frequency of alcohol intake among “consumers” was categorized as daily, weekly or monthly, based on their response on (i) total drinks per month consumed, (ii) drinks per week and/or (iii) drinks per month. For minors, the total number of individuals that reported having consumed an alcoholic drink at some point in life (and not just few sips) was reported.

As in the case of alcohol consumption, smoking habits questionnaire(s) were also based on volunteer’s age. In particular, for adults questionnaires used were from the NHANES [12] and BRFSS [13] studies; for minors from the Youth Behavior Survey [17], NHANES [12] and the European School Survey Project on Alcohol and other Drugs [18]. Volunteers were grouped into (i) current smokers, if they responded that they had smoked the past month, (ii) ever smokers, if they had smoked at any point in their life, and (iii) non-smokers, if they had never smoked. Frequency of smoking, among current smokers, was also recorded as “daily” or “sometimes”. Among minors, the question referred as to whether they had ever tried to smoke (aged up to 19 years).

Weight status was evaluated according to BMI. Body mass Index (BMI) is defined as the weight (in kg) over height (in meters squared). Cut-offs used (all in kg/m²) for assessment are widely used and are the following (adults): < 18.5, underweight; 18.5–24.9, normal weight; 25–29.9, overweight; > 30, obese.

Physical activity has a well-known role as a health determinant hence the aim was to assess physical activity levels in all ages. Questionnaires on physical activity were modified based on age groups as per a priori knowledge, including (i) ≥2- < 12 years old of the questionnaire was based on questions from the NHANES survey [22] and Preschool-aged Children Physical Activity Questionnaire (Pre-PAQ Home Version) [23] (ii) ≥12- < 18 years old, the International Physical Questionnaire – Adolescents (IPAQ-A) [24], (iii) ≥18 years- < 65 years old the IPAQ short form was used [25] and (iv) for ≥65 years of age a modified version of the IPAQ has been suggested [26]. Preliminary results reported in this study include level of physical activity as perceived by the adult volunteers (sedentary, low, moderate and active) or by the primary care giver if the volunteer was < 12 years old.

Information about medical history for disease prevalence among the Greek population, related medical treatment and insurance coverage were collected. The synthesis of this questionnaire was based on the National Health Survey, NHANES [12], ARCHES [20], and the Million Women Study [27]. The definition of clinical investigated outcomes was based on the International Classification of Diseases (ICD)-10th version, recorded by experienced study investigators. Diabetes was defined as fasting blood glucose> 125 mg/dl or if on diabetic medication; dyslipidemia if total triglycerides> 150 mg/dl and/or total cholesterol> 200 mg/dl or on lipid-lowering medication; hypertension as average blood pressure greater or equal to 140/90 mmHg, or on antihypertensive treatment.

Further details on specific disease states (hypertension, dyslipidemia, diabetes) with specific questionnaires [20], were collected once the volunteer declared as having such a condition. In particular, data on prevalence of Chronic Obstructive Pulmonary Disease (COPD) was obtained using the COPD Population Screener [28] and Asthma using the questionnaire from the Hellenic Thoracic Society (for adults), and the Greek version of the questionnaire International Study of Asthma and Allergies in Childhood (ISAAC) [29] (minors 6–18 years old). Following a literature review the Rose Questionnaire for Angina [30] and the Edinburgh Claudication Questionnaire were used in HNNHS [31].

Additional the types of questionnaires used in the study can be viewed in the Additional files 1 and 2 and they included information on breastfeeding, drug and supplement use, memory impairment (≥45 years old), eating habits and behavior (as previously reiterated), sleeping habits, data on depression, stress (acute and chronic) & health locus of control, gestational & child- birth related questions, environmental exposure, functional gastrointestinal disorders, vitamin D intake status & sun exposure, and economic crisis, to acquire adequate and substantial information on the population’s exposures and risks. Details for each of these questionnaires will be provided upon analysis.

HNNHS also included physical examination (temperature, spirometry, blood pressure, etc.), anthropometry (weight, height, waist and hip circumference, body composition, and grip strength), and several blood tests (glucose, HbA1c (diabetics), insulin, total lipid profile, thyroid hormones, thyroglobulin, PTH, complete blood count, folic acid, iron, ferritin, B₁₂, 25OH-vitamin D, creatine, urea, albumin, total protein, ALT, AST, bilirubin, uric acid, calcium, magnesium, manganese, selenium, hs-CRP, cortisol, and heavy metals, namely As, Cd, Co, Hg, Mo, Pb, Pt, Sb, W, Zn, Ce, La, Th, U) in a subsample of the population, to examine correlations with various health indices in later analyses (Additional file 1: Table S3).

The sample was distributed in all different regions of Greece (Table 1). 47.2% was in the region of Attica, 18.5% Central Macedonia, and the rest of the sample being scattered through various regions of Greece (1.3% Epirus, 4.2% Eastern Macedonia and Thrace, 3.1% Peloponnese, 2.2% Western Macedonia, 5.2% Thessaly, 2.3% Central Greece, 4.8% Western Greece, 5.7% Crete, 1.1% Ionian islands, 2% North Aegean and 1.9% South Aegean).

The total number of participants is 4574 volunteers of which 1943 were males and 2629 females. Table 2 shows distribution per gender, age and socioeconomic parameters. Age distribution was representative of the 2011 Census, with 19% (N = 869) of the sampled population being 0–19 years old, 67% (N = 3064), 20–64 years old, and 14% (N = 639) were ≥ 65 years old. Marital status was as follows: 40.6% of the population was unmarried (43.3% males and 38.5% females), 48.4% married (51.4% males and 46.4% females) and 0.1% having a cohabitation agreement 6.2% were widowers (2.2% males and 9.2% females), 3.8% divorced and 0.7% separated.

Educational level greatly varied with approximately 32% having a University degree or greater, 7.1% had completed secondary education. Approximately 17% of the population had limited to low education, 27.1% completed lyceum (12 years of schooling), 5% technical secondary school and 8.3% private post-lyceum college. A large percentage of the population (78.3%; 77.8% males and 78.8% females) reported having public health insurance whereas only 4.3% had private insurance and 8.9% both types. A total of 8% males and 6.2% females were not insured (Table 2).

In terms of net monthly income (Table 2), 13,5% had low income (<€300–850), 11.4% had €851–1050, approximately 18% had moderate high income (€1051–1500), 10.6% had €1501–1900, 9.1% had €1901–2400, and 10.7% had high income (€2401–3800 and > €3801).

Sample’s self-reported weight status in total by age group (> 20 years old) and gender based on Body Mass Index (BMI) can be found in Table 3. A total prevalence of 47.5% of the adult population was overweight (32%) and obese (15.5%), with the prevalence increasing with age in both genders. A significant body weight status difference was found in each age group, with males having a higher prevalence of overweight compared to females (p < 0.001) in all age groups.

Frequency of alcohol consumption among adults was 72.4% (Table 4), with approximately 7% reporting daily consumption, 33% weekly and 60% on a monthly basis. A significant greater percentage of males reported of being alcohol consumers than females (81.1% compared to 67%, respectively; p < 0.001) and being more frequent alcohol consumers as well (p < 0.001). Among minors (12 to 19 years of age, inclusive), 111 out of 340 individuals (32.6%) reported as having consumed an alcoholic drink at some point before, and not only a few sips (Table 4). No significant differences were found between genders among minors in alcohol consumption (p = 0.121).

Smoking frequency in the total population among adults and minors, per gender, is being shown in Table 5. Approximately 34% of the population were current smokers, whereas 50.9% reported on having smoked at some point in their life. Significant gender differences were found in both cases with a higher proportion of males reporting to have smoked (59% compared to 44%) or of being current smokers (38.3% compared to 30.8; p < 0.001 for all). Among current smokers 87.3% reported to smoke daily with a borderline difference found between genders (p = 0.046). A total of 22% of minors (up to 19 years of age, inclusive) reported of having tried to smoke at some point. No significant gender differences were found (p = 0.229).

Preliminary results of physical activity level were self-reported as sedentary, low activity, moderately and very active (Table 6). The highest proportion of the population being very active was in young children (2–12 years old, 68.6%) and among adolescents (48.5%). Twenty – 5 % (25%) of adults aged 18–65 and > 65 years old reported being very active whereas 20% of the elderly (> 65) reported of having a completely sedentary lifestyle.

In Table 7, the prevalence of various chronic diseases is presented in total and per age group (20–39, 40–64, and 65+) in adults. In each category, gender specific rates can also be viewed. The highest prevalence (16.7%) was reported for hyperlipidemia (increased cholesterol or triglycerides), with prevalence increasing in both genders with age (Tukey’s test p < 0.001 between groups). The same pattern was found for hypertension with the prevalence mounting to 56% (51.2% in males, 61% in females; p < 0.05) in the elderly compared to 1.7% in adults aged 20–39 and 17.3% in the 40–65 age group (Tukey’s test not significant). Accordingly, age patterns were seen in all CVD (CHD, angina, MI, heart failure, arrhythmia and stroke), with significant age group differences found only in heart failure (Tukey’s test p = 0.014 for 65+ compared to 20–39 years). Diabetes prevalence and osteoporosis was also considerably higher in the older age group (16,8%) compared to 3.8% in total population and 16.2% compared to 5.4%, respectively. Only osteoporosis was significantly different between age groups (p < 0.001 for 65+ and 20–39 and 40–64). The prevalence of thyroid disease was high in all age groups, especially in females and significantly different between the 65+ and 20–39-year-old age groups (Tukey’s test p = 0.026). A significant difference was also found in cancer prevalence between the older and younger adult age groups (Tukey’s test, p = 0.033).

Significant gender differences were found in hyperlipidemia, arrhythmia, cancer, thyroid disease, osteoporosis, arthritis/rheumatoid arthritis, irritable bowel syndrome, depression, and chronic stress, with females having a significantly higher proportion in each one of them. Prevalence of asthma and cancer was also higher in females, more specifically in the 40–64 age group (4.8% vs. 1.7%; p < 0.05 and 3.7% vs. 0.6%; p < 0.01, respectively). Gender difference was also found in CHD with males having a higher prevalence in the total adult sample and in the older group (p < 0.001, for all). The prevalence of MI did not differ in the total sample but was significantly higher in males over 65 years old than females in the same age group (9.1% vs. 1.9%; p < 0.001). Diabetes mellitus was significantly higher in males aged 40–64 years old than females of the same age group (5.9% vs. 2.7%; p < 0.01).

Due to the cross-sectional nature of the study, no causal relationships can be formulated. Also, the data presented and analyzed in this first report are from reported data. However, experienced field investigators checked the data and recorded clinical outcomes based on ICD-10th version codes. Furthermore, sensitive personal questions, were self-completed to decrease reporting bias. All clinical outcome data were further cross-checked with other related questions, ie, medications, in order to accurately code the participants and decrease misclassification. Reporting of data in more depth and comparison with other past small, non-nationally representative surveys in Greece are beyond the scope of this first methodological publication and will be described elsewhere.

Health surveys as HNNHS can reveal target groups in need for prevention strategies according to educational level, employment and marital status, area of residence in a subnational level, and health behavior [40, 44]. HNNHS, is the first national representative study performed in Greece to assess nutrition and health status of the population including all age groups. Questionnaires used were constructed after performing an extensive literature review and based on other validated questionnaires that have been used in other large national studies and in the Greek population. Another strength is the synergistic action of multiple health care specialists in study design, filed work and data analysis. Furthermore, the use of the especially designed computer software, CAPI, increases reliability of collected data, since it reduces response bias, misclassification and volunteer burden. Measurements, clinical assessment and blood tests performed in a subsample of the population will be used to further validate the preliminary results presented here.